Fastino is real, GLiDE 1.2's 71.5 reproduces from the index's own dataset, and it is cheaper than Jev - but the public-versus-private gap turns out to be a model-size artifact, not a Cloudflare tell
The question
"Who is Fastino, and does GLiDE 1.2 hold up as a decision-model alternative to Jev on contamination-resistant evidence?"
Asked because [[2026-10-24-cloudflare-clef-clairic-vs-typesafe-jev]] went hunting a Cloudflare threat and found an unfamiliar name at the top of the independent board instead. Its own closing line: the accuracy leader "is Fastino GLiDE 1.2 at 71.5 and 93ms, eleven points above Jev, and RDCO has never looked at it." A folded-in second scope asks whether the public-versus-private score gap that brief read as Clef's contamination signature is general across all 228 reproductions.
Method and evidence labels
Same four labels as the parent brief: (vendor) for a vendor's own claim, (vendor-measured) for a performance number the vendor measured, (independent) for a party selling none of these models, (U) for unverified or unpublished.
One methodological upgrade matters. Rather than reading the Decision Index's rendered leaderboard, this pass downloaded the index's own published data files (data/index.json, data/own.json, data/methodology.json, generated 2026-10-26T15:30:58Z) and computed the statistics over all 229 entries directly. Every board number below is read off that dataset, not off a screenshot, a tooltip, or a press summary. The score-field mapping was pinned by reconciliation: public_skill reproduces the parent brief's Clef 61.7 and balanced_skill reproduces its Clef 54.9, which confirms public_skill is the PUBLIC ONLY toggle and balanced_skill is the FULL SCORE private column.
What we already know (from the vault)
- The dedupe gate holds. A fresh
qmdpass returns zero vault coverage of Fastino or GLiDE outside the parent brief's two passing mentions. This is a clean gap. - The parent brief's Jev-stays recommendation rests on exactly the claim this brief tests. [[2026-10-24-cloudflare-clef-clairic-vs-typesafe-jev]] kept Jev because "it leads on the only independent evidence that is contamination-resistant." If GLiDE outranks Jev on that same private column, the stated reason does not hold as written.
- Two reusable gotchas carried in from the parent brief, both re-confirmed in the raw data. The
#column is a 0.9-point tie band, not a rank: Jev 1.14 carriesrank_tie: 10andtied: truewhile its actualrank_vis 42 of 229. And the FULL SCORE / PUBLIC ONLY toggle reverses orderings, which is why the field mapping above had to be pinned before anything else. - RDCO's integration is two constants deep.
/Users/ray/.claude/skills/ideas/jev_judge.pyis 301 lines, standard-library only, hardcodingAPI_URL = "https://api.typesafe.ai/v1/systemone"andMODEL = "jev-latest". The parent brief pre-registered a--providerflag test on exactly this basis. - The standing default is [[feedback_jev_for_typed_set_problems]] (founder, 2026-10-11). The licence-boundary posture that the Fastino licence has to be read against is [[2026-05-05-personal-license-boundary]].
What the web says
- Fastino is a real, funded, operating company with a live self-serve product. Primary surfaces verified first-hand this pass:
fastino.ai("Fastino Labs is an applied research lab building open weight language models and tools for production AI", footer "Fastino Inc."),docs.fastino.aiwith a published rate card, a self-serve account and API-key flow atagent.fastino.ai, and weights athuggingface.co/fastino/GLiDE. Funding is secondary-sourced: a $7M pre-seed led by Insight Partners and M12 (Microsoft's venture arm) around November 2024, and a $17.5M seed led by Khosla Ventures announced around May 2025, roughly $25M total. The amounts come from aggregators and one law-firm announcement rather than a company page, so treat the figures as (U) while treating company existence as settled. - GLiDE 1.2's 71.5 reproduces exactly from the index's own dataset.
balanced_skill = 71.52,public_skill = 71.99,balanced_raw = 78.89,rank_v = 1 of 229,rank_tie = 1,tied = false, coverage 1.0 with all 110,201 requests completed and zero errors (independent). Against Jev 1.14 atbalanced_skill = 60.24, that is +11.28 points on the contamination-resistant column. - The win is broad, not a single-benchmark spike. GLiDE beats Jev 1.14 in all five scored categories on the skill-normalized view (knowledge .606 against .571, language .755 against .612, retrieval .722 against .635, tools .880 against .738, arts .626 against .392) and on both halves of the private score (same-skills 68.57 against 59.28, new domains 59.66 against 48.26).
- GLiDE is also one of the best-calibrated models on the board, where Jev is mid-pack. Expected calibration error 0.0204, 4th of 229, against Jev's 0.0799 at 127th of 229 and Clef's 0.0395 at 33rd (independent). Calibration is scored on a 1-in-6 sample across 32 benchmarks, and the index's own note says "Rows an entrant trained on count as wrong here too." For a model class whose entire selling point is a trustworthy probability, this is the most decision-relevant number in the brief after the Full score.
- It is cheaper than Jev, and the price is primary-sourced twice.
docs.fastino.ai/pricinglists GLiDE inference at $0.04 per 1M input tokens, $0 output, anddocs.fastino.ai/inference/systemonerepeats it in prose: "GLiDE costs $0.04 per 1 million input tokens. Output tokens are free ($0)" (vendor). Jev is $0.042. That is 4.8% below Jev, which is to say immaterial, but it removes price as any argument for staying. - The API is Jev-shaped to a degree that goes beyond Cloudflare's compatibility claim. The endpoint is
POST https://api.fastino.ai/v1/systemonewithmodel: fastino/GLiDE, taking astateplus named typedquestionsand returninganswers; the docs say explicitly that GLiDE "does not use/v1/chat/completions,messages, or a GLiNERschema." The primitives are Noul, Choice and Score, the same three Jev exposes. Limits are 65,536 tokens per rendered prompt and up to 4 images natively, against Jev's images gated behind aTypeSafe-Danger-Zone-Imagesheader. - The licence is permissive for RDCO and restrictive only above $10M revenue. The
LICENSEfile at the weights repo is the Fastino Community License (Copyright 2026 Fastino Inc., applying to releases on or after 2026-09-30). It grants use, copy, modify, merge, publish, distribute and sublicense free of charge, with the material condition that an entity whose annual revenue exceeds $10,000,000 US dollars must obtain a separate written commercial agreement for production commercial use. The Qwen3.8-27B base stays under Apache 2.0. - The weights are obtainable but not anonymously. The model card's install line is
hf download fastino/GLiDEand notes it "needs a token with read access to the fastino org." That is a weaker openness posture than Clef's unconditional Apache 2.0 repos, and it means "open weights" here is closer to "community-licensed weights behind an org read token" (U on whether the repo is formally gated).
The 93ms is a median on a datacenter card, and the tail is the real story
The parent brief carried "71.5 and 93ms" as a pair. The first number survives scrutiny. The second needs a correction that changes its meaning.
| Source | Who | Median | p95 | Mean | Rig |
|---|---|---|---|---|---|
| Decision Index | independent | 92.9ms | 1,121.1ms | 284ms | 1x NVIDIA RTX PRO 6000, local process |
| Fastino model card | vendor-measured | 84ms (RTX PRO 6000), 48ms (B200), 78ms (H100) | p99 2,709ms (RTX PRO 6000), 1,086ms (B200) | not stated | local, first 300 requests |
| Jev 1.14, same index | independent | 196ms | 296.8ms | not stated | hosted HTTPS round trip from the lab |
Three readings follow, and only the first is the one in circulation.
- On the median, GLiDE is about twice as fast as Jev. 92.9ms against 196ms.
- On the 95th percentile, Jev is about 3.8x faster. 297ms against 1,121ms. GLiDE's mean of 284ms exceeds Jev's median, which is the signature of a heavy right tail rather than a slow model.
- Neither comparison is like-for-like, and the index says so. GLiDE runs in-process on a graphics card; Jev is measured over the public internet, with the index's own note reading "Hosted API, network round-trip from our lab... Not comparable to the on-card single-process figures." Fastino's own card adds a 25-35s cold start and a 30-60s weight load, and reports a single 39k-token question taking about 13s warm.
The tail is not an accident, and it is not GLiDE-specific. The index entry GLiDE 1.2 replaced is named decision-27b-ckpt-nothink, and every one of the top ten models on the private column carries a p95 between 674ms and 2,245ms while Jev's hosted p95 sits at 297ms. The models that win the contamination-resistant column are adaptive-compute models that think longer when uncertain, and they pay for it in tail latency. For RDCO's /ideas batch judgments that is irrelevant. For anything on a request path it is the whole question, and it is a trade the "93ms" headline hides.
The gap question: general, but it is a size effect, not a category-wide contamination tell
This is the scope that got folded in, and it comes back with a more useful answer than expected. Measuring public_skill - balanced_skill across all 229 entries:
| Statistic | Value |
|---|---|
| Mean gap | +4.13 |
| Median gap | +2.90 |
| Standard deviation | 6.29 |
| Range | -5.94 to +24.91 |
| Positive (public flatters) | 161 of 229 (70%) |
| Negative | 68 of 229 (30%) |
| Gap above Clef's +6.79 | 63 of 229 |
So the first-order answer is yes, general: on 70% of the board the public column reads higher than the private one, by a median of nearly three points. But stratifying by parameter count changes what that means.
| Size band | n | Mean gap | Median gap | Share positive |
|---|---|---|---|---|
| Under 2B | 58 | +7.24 | +4.60 | 81% |
| 2-6B | 58 | +5.71 | +4.45 | 79% |
| 6-15B | 31 | +4.89 | +3.23 | 87% |
| 15-40B | 78 | +0.44 | -0.11 | 50% |
Pearson correlation between log-parameters and gap: r = -0.417. By training method: full fine-tunes average +5.84 (n=127), LoRA adapters +2.42 (n=44), pure inference techniques -0.70 (n=26). The pattern is coherent: the entrants that lose the most ground on held-out tests are small models fully fine-tuned on the visible suite.
Three consequences, in order of how much they change RDCO's behaviour:
- At the size class where this decision actually lives, the gap is zero on average. In the 15-40B band that contains Jev, Clef, GLiDE and every serious contender, mean gap is +0.44 and only half the entries are positive at all. Jev's gap is +0.46 and GLiDE's is +0.47, both unremarkable for their class.
- Clef's +6.79 therefore survives as an outlier, but only once you control for size. It ranks 8th largest of the 78 entries in its own size band, roughly the 90th percentile, and
clef-omniis the single worst in that band at +11.50. The parent brief's contamination read on Clef stands. Its implied generalization - that the gap is "the signature of public-suite contamination" as a category property - does not, because the category-wide gap is explained better by model size than by vendor behaviour. - The honest reading rule is a conditional one, and it is cheap to apply. A published public-benchmark score for a sub-6B decision model should be discounted by roughly 5 to 7 points before it is compared against anything. A public score for a 27B-class model needs no size discount, so a large gap there is genuinely informative about that specific entrant. This is the reusable finding, and it is more durable than any single model ranking.
One caution this brief cannot resolve: a size-dependent gap is consistent with contamination and also consistent with small models simply generalizing worse. The index's own framing is contamination ("Anyone can see them, and train on them, so they no longer count toward the Full score"), and it has removed at least one entrant for provably training on test items (reflex-4b, dropped because its adapter "was trained on 800 MMLU-Pro test items"). But capacity and contamination predict the same correlation, and nothing in the published data separates them. Calling the gap a contamination signature is (U) as a mechanism, even though it is settled as a measurement.
Convergences and contradictions
- Everything about GLiDE's score converges, which is itself the finding. The index dataset says 71.52 at rank 1; Fastino's homepage chart says 71.5 against "Jev 1.14, 60.2"; the model card says "full score of 71.5 on the Decision Index 0.4.1" and "ranked #1 of 229 models." The vendor is citing the independent board rather than publishing its own suite, and the board's raw data agrees with the vendor's citation of it to two decimal places. There is no vendor-versus-independent contradiction to adjudicate here, unlike the Cloudflare case.
- The provenance of GLiDE's run is the one real gap, and it cuts against over-trusting rank 1. The entry's metadata reads
"pr": 255, "new": true, "replaces": "decision-27b-ckpt-nothink", so it arrived as a submission. The index's documented practice for author-submitted runs is independent partial reproduction before adoption: its changelog records adopting one entrant "after a 3,000-row reproduction matched it bit for bit" and another "after our latency-v1 run agreed with it." No such adoption note exists for PR #255 in the published methodology data, and that file is still at editionv0.2.1while the board is at 0.4.1, so the provenance document lags the ranking it should document. Mitigating this: the entry records a pinned weights revision (a7e45896be7c96ba3354891da790c5ec99c30b6c), full coverage, and zero errors, so the run is at least precisely specified and re-runnable by anyone. Whether the index re-ran it is (U). - The parent brief's stated reason for keeping Jev no longer holds, but its conclusion does. "Jev leads on the only independent evidence that is contamination-resistant" is now false: Jev is 42nd of 229 on that column, 41 entries outscore it on the Full score (up from 40 at the 2026-10-24 reading), 47 outscore it on the public column, and 29 beat it on both private halves at once. Jev's defensible claims are narrower and still real: a tight latency tail, a production hosted endpoint, and the fact that it is already wired into RDCO.
- Minor source conflicts, flagged rather than resolved. Context window is 65,536 tokens in the docs and "up to 40,000 tokens per question prompt by default" on the model card. Parameters are "26B params" in the Hugging Face header, 27.8B in the index's served-params field, and a 27B base in the card text. The card says "Inference provider: Fastino" while the Hub widget says the model "isn't deployed by any Inference Provider." None of these change the decision; all of them are the kind of thing that would be quoted wrongly later.
Synthesis for RDCO
Fastino clears the reality bar comfortably, and GLiDE 1.2 clears the evidence bar on every axis this brief could measure. The company has a published rate card, a self-serve key flow, versioned docs with cookbooks, a licence file with a real revenue threshold, and institutional money behind it. The model is first of 229 on the index's contamination-resistant column by 1.3 points over the runner-up and 11.3 over Jev, it wins all five categories and both private halves rather than spiking on one, it is fourth of 229 on calibration where Jev is 127th, and it costs $0.04 per million input tokens against Jev's $0.042. The licence matters less than it looks: production commercial use needs a separate agreement only above $10M annual revenue, and RDCO is nowhere near that, so under [[2026-05-05-personal-license-boundary]] the weights are usable today and the threshold is a problem worth having. The practical catch is smaller than the licence: downloading the weights needs a token with read access to the Fastino organisation, so this is community-licensed-behind-a-token rather than Clef's unconditional Apache 2.0.
The evidence does not, however, settle that GLiDE beats Jev on RDCO's workload, and the brief should not pretend otherwise. Everything above rests on one leaderboard, one pull-request-submitted entry, and no published note that the index re-ran it. There is no independent replication of Fastino's run by a second party, no hosted-to-hosted latency comparison against Jev, no published free tier, rate limits or service-level agreement, and no training-data disclosure. The measured latency profile is genuinely worse than Jev's where it counts for anything interactive: GLiDE's p95 of 1,121ms against Jev's 297ms, with Fastino's own card reporting a p99 of 2,709ms and a 25-35 second cold start. "Rank 1 on a leaderboard at a 5% lower list price" is a strong reason to run a test. It is not a product recommendation, and the gap between those two things is exactly what the parent brief's pre-registered test exists to close.
The gap scope returns the most reusable result of the night, and it partly corrects the parent brief. The public-versus-private spread is general across the board - 70% of entries, median +2.9 points - but it is driven by model size, not by vendor conduct: mean +7.24 under 2B falling to +0.44 at 15-40B, with r = -0.417 against log-parameters and full fine-tunes averaging more than twice the gap of LoRA adapters. So the reading rule RDCO should adopt is conditional rather than blanket. Discount a published public-benchmark score for a small decision model by roughly five to seven points before comparing it to anything; apply no size discount at 27B class, which makes a large gap there informative about that entrant specifically. On that rule Clef's +6.79 is still an outlier for its class at the 90th percentile and the parent brief's contamination read on Cloudflare survives, while its wider implication that the whole category is contaminated does not. And the mechanism itself stays open: capacity limits and benchmark contamination predict the same size correlation, and the published data cannot separate them.
For the standing default, the recommendation is unchanged but the reasoning has to be rewritten, and the test just got cheaper to run. Jev should stay the default under [[feedback_jev_for_typed_set_problems]] for three reasons that are all still true: it is wired and working, its latency tail is the tightest of any serious contender, and no alternative has been measured on RDCO's own labelled sample. The reason that must be retired is the one the parent brief leaned on, that Jev leads the contamination-resistant evidence. It does not; it sits 42nd. What has changed materially is the cost of finding out. Fastino's endpoint is POST /v1/systemone taking state plus typed questions, exposing Noul, Choice and Score, which is the same shape TypeSafe serves at api.typesafe.ai/v1/systemone. The --provider {jev,clef,clef-flash,glide} flag the parent brief pre-registered is still a two-constant change plus a credential for GLiDE, and GLiDE should now be the first alternative arm run rather than the last, because it is the only one that leads Jev on held-out data, on calibration, and on price simultaneously. The acceptance bar from the parent brief holds verbatim: no published benchmark counts, including the Decision Index, and GLiDE replaces Jev only if it wins on agreement against RDCO's own withheld labels at acceptable measured latency from the Mac mini.
Why this is in the vault
[[2026-10-24-cloudflare-clef-clairic-vs-typesafe-jev]] kept Jev as RDCO's default for typed-set judgments on the explicit ground that Jev led the contamination-resistant evidence, and that ground is now measurably wrong (Jev is 42nd of 229 on the private column; 41 entries beat it). This brief supplies the corrected basis for the [[feedback_jev_for_typed_set_problems]] default, names GLiDE as the first arm the pre-registered jev_judge.py provider-flag test should run, and gives RDCO a size-conditional rule for reading every published decision-model benchmark from here on, replacing the blanket "discount the public column" heuristic the parent brief implied.
Open follow-ups
- Does the Decision Index independently re-run author-submitted entrants, and is there an adoption note for PR #255? The published methodology file is still edition v0.2.1 while the board is 0.4.1, so the provenance document for the edition that puts GLiDE at rank 1 does not cover it. The index's GitHub repo (
apolinario/decision-index) ships the reproduction scripts, so this is answerable rather than merely open. - Is the size-dependent public-versus-private gap contamination or capacity? Both predict r = -0.417. Separating them would need per-benchmark overlap analysis against each entrant's training corpus, and would settle whether "leaderboard hygiene" or "small models generalize worse" is the right frame for the whole category.
- What is GLiDE's latency from RDCO's own Mac mini over HTTPS, and does Fastino publish a free tier, rate limits or a service-level agreement? Every GLiDE latency figure in existence is on-card and local; every Jev figure is a hosted round trip. The only comparison that decides anything for RDCO is hosted-to-hosted on its own payloads, and no vendor pair has ever been measured that way.
- Does Fastino disclose GLiDE's training data, and does the Fastino Community License survive a change of control? Cloudflare at least confirmed its training data was undisclosed; Fastino has not been asked. The $10M revenue clause is reassessed quarterly, and a licence written by a Series-A company is worth reading for what happens if that company is acquired.
- Would a GLiDE fine-tuned on RDCO's own labelled idea-bank sample beat all three off-the-shelf options? Fastino sells managed fine-tuning at $0.07-$0.13 per GPU-minute plus a Tinker-compatible training path, which is a capability neither TypeSafe nor Cloudflare offers self-serve. If the pre-registered test shows any provider close to RDCO's accuracy bar, tuning rather than switching may be the cheaper move.
Related
- [[2026-10-24-cloudflare-clef-clairic-vs-typesafe-jev]] - the parent brief; this one corrects its contamination generalization and retires its stated reason for keeping Jev while upholding the decision
- [[2026-10-20-technically-whats-a-decision-model]] - the vault's conceptual primer on the decision-model class; still the right place to start on what Noul, Choice and Score are
- [[2026-09-27-alphasignal-system-one-models-jev-laya-clm8b]] - logged the arrival of open Jev rivals and called for a head-to-head; GLiDE is the strongest instance of what that note anticipated
- [[2026-09-23-every-jev-usage-guide]] - source of the vault's $0.042 Jev figure, which GLiDE's $0.04 now undercuts
- [[2026-09-15-every-typesafe-jev-vibe-check]] - opened the Jev due-diligence thread this brief extends
- [[2026-09-28-indy-dev-dan-10-levels-of-jev]] - the Jev-guard hook pattern a provider flag would make vendor-neutral
- [[2026-10-19-indy-dev-dan-jev-news-curator-agent]] - source of the "sub-half-second" shorthand, which GLiDE's p95 of 1,121ms would break
- [[2026-10-22-automode-classifier-channel-text-injection-surface]] - where the typed-values-shrink-the-injection-surface argument applies; Fastino ships a prompt-injection cookbook for GLiDE
- [[2026-05-05-personal-license-boundary]] - the founder's licence-boundary position memo, against which the Fastino Community License $10M revenue clause reads as usable today
- [[feedback_jev_for_typed_set_problems]] - the standing default this brief re-grounds and leaves in place
Sources
Independent measurement (primary data, downloaded and computed over this pass)
- https://multimodalart-jev-decision-index.static.hf.space/data/index.json - Decision Index 0.4.1,
generated_utc 2026-10-26T15:30:58Z, editionrelease-v4, 229 entries, 110,201 requests, 1x NVIDIA RTX PRO 6000,jev_version jev-1.14.0. GLiDE 1.2:balanced_skill 71.52,public_skill 71.99,balanced_raw 78.89,rank_v 1,tied false, latency median 92.9 / p95 1121.1 / mean 284, ECE 0.0204, coverage 1.0,meta.pr 255,replaces decision-27b-ckpt-nothink,served_checkpoint a7e45896be7c96ba3354891da790c5ec99c30b6c. Jev 1.14:balanced_skill 60.24,public_skill 60.70,rank_v 42,rank_tie 10, latency median 196 / p95 296.8, ECE 0.0799. Clef:public_skill 61.71,balanced_skill 54.92,rank_v 69. All gap, size-band, correlation and calibration-rank statistics in this brief were computed from this file. - https://multimodalart-jev-decision-index.static.hf.space/data/methodology.json - edition v0.2.1 (lags the 0.4.1 board); the entrant-classification rules, the
reflex-4bremoval for training on 800 MMLU-Pro test items, and the author-submission adoption notes ("adopted after a 3,000-row reproduction matched it bit for bit") - https://multimodalart-jev-decision-index.static.hf.space/data/own.json - the private same-skills and new-domains per-area breakdown; Full score weights (
same_skills 0.6,new_domains 0.4,public 0) - https://huggingface.co/spaces/multimodalart/jev-decision-index - the rendered board; footer "Not affiliated with TypeSafe AI"
- https://github.com/apolinario/decision-index - the index's reproduction scripts (named in
methodology.json, not inspected this pass)
Fastino (vendor-primary)
- https://huggingface.co/fastino/GLiDE - model card: "GLiDE is a modified version of Qwen3.8-27B";
hf download fastino/GLiDE"needs a token with read access to the fastino org"; self-reported "full score of 71.5 on the Decision Index 0.4.1", "ranked #1 of 229 models"; median 84ms RTX PRO 6000 / 48ms B200 / 78ms H100, p99 2,709ms / 1,086ms; 25-35s cold start, 30-60s weight load; 40,000-token default per question - https://huggingface.co/fastino/GLiDE/raw/main/LICENSE - Fastino Community License, Copyright 2026 Fastino Inc.; applies to releases on or after 2026-09-30; permissive grant conditioned on a separate written commercial agreement for production commercial use by entities with annual revenue above $10,000,000, reassessed quarterly; Qwen3.8-27B stays Apache 2.0
- https://docs.fastino.ai/pricing - GLiDE $0.04 per 1M input tokens, $0 output; GLiNER-2.5-Decide $0.03; training config jobs $0.07-$0.13 per GPU-minute
- https://docs.fastino.ai/inference/systemone -
POST https://api.fastino.ai/v1/systemone,stateplus named typedquestionsreturninganswers; "GLiDE does not use/v1/chat/completions,messages, or a GLiNERschema"; the $0.04 / $0 restatement - https://docs.fastino.ai/concepts/glide-capabilities - 65,536 tokens per rendered prompt; up to 4 images; confidence semantics and the warning to "tune action and fallback thresholds on representative data rather than assuming a universal cutoff"
- https://docs.fastino.ai/llms.txt - the documentation index; Noul / Choice / Score primitives; prompt-injection, agent-gating and ticket-triage cookbooks; Auto Train as private access
- https://www.fastino.ai/ - "Fastino Labs is an applied research lab building open weight language models and tools for production AI"; "GLiDE is our state of the art decision model, ranked #1 on Hugging Face's Decision Index"; chart showing GLiDE 71.5 against "Jev 1.14, 60.2"; no pricing or latency figures on the page
Secondary, carried as (U)
- Funding: $17.5M seed led by Khosla Ventures (~May 2025) and a $7M pre-seed led by Insight Partners and M12 (~November 2024), ~$25M total, per seedtable.com, vcbacked.co, hpcwire.com/aiwire and wsgr.com. No company-published funding page was found, and two aggregators disagree on the total ($25M against $24.6M).
- No independent third-party review or replication of GLiDE by any party other than the Decision Index was found this pass. Fastino's "up to 99x faster than standard LLMs" and sub-$100k training-hardware claims are vendor marketing and were not tested.
Vault paths cited
- /Users/ray/rdco-vault/06-reference/research/2026-10-24-cloudflare-clef-clairic-vs-typesafe-jev.md
- /Users/ray/rdco-vault/06-reference/2026-10-20-technically-whats-a-decision-model.md
- /Users/ray/rdco-vault/06-reference/2026-09-27-alphasignal-system-one-models-jev-laya-clm8b.md
- /Users/ray/rdco-vault/06-reference/2026-09-23-every-jev-usage-guide.md
- /Users/ray/rdco-vault/06-reference/2026-09-15-every-typesafe-jev-vibe-check.md
- /Users/ray/rdco-vault/06-reference/2026-09-28-indy-dev-dan-10-levels-of-jev.md
- /Users/ray/rdco-vault/06-reference/2026-10-19-indy-dev-dan-jev-news-curator-agent.md
- /Users/ray/rdco-vault/06-reference/research/2026-10-22-automode-classifier-channel-text-injection-surface.md
- /Users/ray/.claude/skills/ideas/jev_judge.py - the two constants a
--provider glidearm would change