You're running ab94174 · shipped 2026-10-22. 120 commits since 4b867cc on 2026-06-23. Each feature below shows when it was first added and last touched — click any hash to see the exact change on Bitbucket. The version badge (bottom-left) always links to the running build.
Feature freshness — per sidebar link: introduced → last update
◎ Overview
7×added 2026-06-23 · 2204358
last update 2026-09-20 · 21e3c0a
Landing page: derive the provider count instead of typing it in
⚡ Router
11×added 2026-06-23 · 4b867cc
last update 2026-09-20 · a3154ca
Jev: fix 29 findings from an adversarial review — Retry-After, fail-closed trimming, and s…
◑ Use cases
16×added 2026-06-23 · 4b867cc
last update 2026-09-21 · 0a4671f
Jev: fold in a third-party explainer video, after checking every claim in it
∞ RSI Lab
17×added 2026-07-08 · e8c62d8
last update 2026-10-22 · ab94174
Architectures: start from Karpathy's microgpt; +27 checked articles across four pages
⇶ Providers
5×added 2026-07-17 · 6ee1884
last update 2026-09-20 · a3154ca
Jev: fix 29 findings from an adversarial review — Retry-After, fail-closed trimming, and s…
🔥 Trending
11×added 2026-07-20 · 3df6d33
last update 2026-10-15 · 28edf34
Trending: catch up 12 weeks of a stalled feed, and add REA with its star count flagged
⌬ Architectures
13×added 2026-06-24 · d76ca41
last update 2026-10-22 · ab94174
Architectures: start from Karpathy's microgpt; +27 checked articles across four pages
🧠 Second brain
3×added 2026-08-22 · dc60a38
last update 2026-08-22 · dfc576a
Artificial Hivemind (arXiv 2510.22954): pluralism caveats across brain, router, RSI
⚖ AI Compliance
5×added 2026-06-26 · 6085244
last update 2026-09-20 · 84269cd
Jev use cases: +16 from the security, compliance and data-engineering analysis lane
💰 Cost Lab
1×added 2026-08-19 · eedca43
last update 2026-08-19 · eedca43
2026 sweep: AI Compliance + Cost Lab sections, verified model & benchmark refresh
⇌ AI Gateway
10×added 2026-07-23 · e0f0c88
last update 2026-07-25 · 5f82597
Gateway: feature Hugging Face + aisuite provider:model compatibility
☁ Cloudflare
4×added 2026-06-25 · 384f31c
last update 2026-09-02 · c8a69aa
Feature Hermes Agent, ponytail, and Cloudflare MCP Portals; cite sources
⬇ Run locally
9×added 2026-06-23 · 8a794d8
last update 2026-09-21 · 0a4671f
Jev: fold in a third-party explainer video, after checking every claim in it
▤ Benchmarks
8×added 2026-06-23 · 4b867cc
last update 2026-10-22 · ab94174
Architectures: start from Karpathy's microgpt; +27 checked articles across four pages
⎙ Benchmark papers
6×added 2026-06-28 · f4ca7f6
last update 2026-10-15 · fb3681b
What's new: stop crediting every Trending commit to Benchmark papers
⬡ Agent harnesses
11×added 2026-06-23 · 4b867cc
last update 2026-09-21 · 0a4671f
Jev: fold in a third-party explainer video, after checking every claim in it
❖ Agent skills
6×added 2026-06-23 · 4b867cc
last update 2026-10-15 · 28edf34
Trending: catch up 12 weeks of a stalled feed, and add REA with its star count flagged
◇ Vector databases
8×added 2026-06-23 · 4b867cc
last update 2026-09-21 · 0a4671f
Jev: fold in a third-party explainer video, after checking every claim in it
⟳ Sources & freshness
25×added 2026-06-23 · 4b867cc
last update 2026-09-05 · a2d7e08
Models + architectures: the Aug/Sep 2026 wave, verified against primary sources
⚙ Settings & sign-in
10×added 2026-06-24 · 7144e4a
last update 2026-07-23 · 47f6c2a
Login required by default + live "What's new" strip and build stamp on the landing
Full timeline — every commit, newest first, back to #1
ab94174
2026-10-22
Architectures: start from Karpathy's microgpt; +27 checked articles across four pages
∞⌬▤
28edf34
2026-10-15
Trending: catch up 12 weeks of a stalled feed, and add REA with its star count flagged
🔥❖
0a4671f
2026-09-21
Jev: fold in a third-party explainer video, after checking every claim in it
◑∞⬇⬡◇🗓
6cefa48
2026-09-20
Jev use cases: dedupe to 85 on exact lane provenance, and fill the thin categories
◑
84269cd
2026-09-20
Jev use cases: +16 from the security, compliance and data-engineering analysis lane
◑⚖
34a7577
2026-09-20
Jev use cases: 62 worked examples, each carrying the question pack you would actually send
◑
a3154ca
2026-09-20
Jev: fix 29 findings from an adversarial review — Retry-After, fail-closed trimming, and six misattributed claims
⚡⇶⌬
6bc832e
2026-09-20
Changelog: regenerate from git history — it had stalled at 94 commits (2026-07-28)
🗓
2fdd8cb
2026-09-20
Jev: wire TypeSafe's System One model into triage, routing and pre-flight screening
⚡◑∞⇶⌬⚖⬇⬡◇🗓
c2274e9
2026-09-06
Architectures: four shipped mechanisms from the arXiv sweep, and Gemma 4
◑∞⌬⬇⎙⬡◇🗓
7ffb4a7
2026-09-06
GPT-6 Astra + Fable 5.1: replace the empty "undisclosed" cells with what is actually published
⌬
01d7534
2026-09-05
NVIDIA: add Qwen3.5, GLM-5.2/4.7 and the Mistral 3 generation; correct the evidence note
⌬
a2d7e08
2026-09-05
Models + architectures: the Aug/Sep 2026 wave, verified against primary sources
◑∞⌬⬇⬡◇⟳🗓
e556370
2026-08-24
Data as markdown: md/ mirrors all 18 catalogs, generated by scripts/export-md.mjs
◑∞⬇⬡◇🗓
dfc576a
2026-08-22
Artificial Hivemind (arXiv 2510.22954): pluralism caveats across brain, router, RSI
⚡∞🧠▤⎙
b675ec8
2026-08-19
Router: add Routing research (/app/router/research) — the 2026 literature, verified
⚡
eedca43
2026-08-19
2026 sweep: AI Compliance + Cost Lab sections, verified model & benchmark refresh
⌬⚖💰▤⟳🗓
88494e2
2026-07-23
Gateway: onboard any OpenAI-compatible provider — bring your key, verified bases
⇶⇌🗓◆
0725042
2026-07-23
Visibility: pin the build badge on-screen; move "What's new" up under the hero
◎🗓
efba639
2026-07-23
Landing: build footer shows the exact running build (APP_COMMIT), matching the sidebar badge
◆
47f6c2a
2026-07-23
Login required by default + live "What's new" strip and build stamp on the landing
⚙🗓
a2a4273
2026-07-23
Gateway: free-tier budget board + auto/free headroom routing + animated strategies
⌬⇌🗓
ca8fa67
2026-07-22
Gateway: OpenAI-compatible /v1 endpoint with quota-aware auto-fallback (OmniRoute-style)
⇌🗓◆
e0f0c88
2026-07-23
Trending ingest 2026-07-23: +8 (diegosouzapw/OmniRoute, stablyai/orca, earendil-works/pi, microsoft/SkillOpt, rtk-ai/rtk, alibaba/open-code-review, evid)
🔥⇌❖🗓
1a5a86c
2026-07-21
Architectures: Kimi K3 + GLM-5.2 deep dives — bespoke diagrams, primary-source verified
⌬🗓
713e655
2026-07-20
Trending: pick the best CURRENTLY-available Groq model at runtime + --check preflight
🔥
eb8a8d5
2026-07-20
Trending ingest 2026-07-20: +8 (AstrBotDevs/AstrBot, Alishahryar1/free-claude-code, lyogavin/airllm, moonshine-ai/moonshine, garrytan/gstack, cjpais/Han)\n\nAuto-curated from github.com/trending by heuristic. Data-only, dedup'd, capped at 8/run.
🔥🗓
3df6d33
2026-07-20
Trending: daily autonomous github.com/trending ingest, curated by best Groq model
🔥🗓
6ee1884
2026-07-17
Providers directory: the inference ecosystem, seeded by Baseten (/app/providers)
⇶
28a0097
2026-07-08
RSI Lab: explicit model picker + loops dropdown (3–15) + "until acceptable" mode
∞
6ccf411
2026-07-08
RSI Lab: algorithms atlas, domain "what-good-looks-like" models + Further reading
∞
6d7b5ff
2026-06-29
Sales: 7 net-new AE + sales-leadership recipes from the Owner.com AI-sales talk
◑
72d46f7
2026-06-29
GTM use cases: 16 new recipes mined from Kyle Norton's "Inside Our AI Sales Stack" (Owner.com)
◑
09b2fb0
2026-06-29
Add 2026 frontier benchmarks (CritPt, HMMT, IMO-AnswerBench, ProgramBench, FrontierSWE, SWE-Marathon, PostTrainBench, MCP-Atlas, Tool-Decathlon) + DeepSWE agent
▤⬡
dcea1f7
2026-06-29
Coding agents + benchmarks: fill the SWE-bench Pro / Multilingual / Terminal-Bench 2.1 / NL2Repo / BigCodeBench gaps and add 5 missing coding-agent harnesses
▤⬡
75ae0ea
2026-06-28
Benchmark radar: arXiv trend intelligence (concept clouds, rising ideas, new-benchmark tracker)
⎙
5d174ea
2026-06-26
Force a fresh deploy snapshot (recipes-40 marker) — Railway dedup was skipping the 40-recipe deploy
◑◆
aba9488
2026-06-26
Router page: unified model directory across all sources + single/compare tester
⚡⟳◆
6085244
2026-06-26
Recipes: add 10 more use cases across Security, Compliance, Sales, Ops (10 → 20)
◑⚖