This website requires JavaScript.
4f37651620
Add Opus 4.8 (Claude Code) v2 runs for both cases (#10 )
Junchao Mao and Claude Opus 4.8
2026-06-13 03:55:13 -07:00
5f6a44aff7
Qwen3.7-Max: use the red-bike pelican variant
Jesse Zhang and Claude Opus 4.8
2026-06-13 11:52:38 +01:00
f21944db77
Add Qwen3.7-Max (Qoder) runs for both cases
Jesse Zhang and Claude Opus 4.8
2026-06-13 11:43:21 +01:00
70b13fcee2
feat: add Kimi K2.7-Code under Kimi Code (#9 )
OHCJ099 and OHCJ099
2026-06-13 17:06:25 +08:00
dc9ad36359
Register Mistral Medium 3.5 × Vibe (Thinking)
Jesse Zhang and Claude Opus 4.8
2026-06-12 23:37:42 +01:00
f4f4530797
feat: add Nex-N2-Pro under Claude Code (#8 )
OHCJ099 and OHCJ099
2026-06-12 17:31:48 +08:00
9b58bac621
Composer 2.5 Cursor "Fast": effort is Default, not a thinking budget
Jesse Zhang and Claude Opus 4.8
2026-06-11 22:40:31 +01:00
df5a53a6b6
Composer 2.5 Fast: label is "Composer 2.5", "Fast" is the effort
Jesse Zhang and Claude Opus 4.8
2026-06-11 21:32:36 +01:00
db5b9beb14
Register Claude Fable 5 × Cursor (High) (#7 )
驱不散的雾 and Cursor
2026-06-12 03:24:48 +08:00
6fbf660b49
Register Composer 2.5 Fast × Cursor (Max) (#6 )
驱不散的雾 and Cursor
2026-06-12 03:24:44 +08:00
44ebf586aa
README: unlock the board at 3 total votes (was 3 per case)
Jesse Zhang and Claude Fable 5
2026-06-11 20:01:08 +01:00
816fe537c3
README: add an AI-agent contribution callout to both languages
Jesse Zhang and Claude Fable 5
2026-06-11 16:11:04 +01:00
0facc860b3
Set Composer 2.5 (Grok Build TUI) effort to Default, not blank
Jesse Zhang and Claude Fable 5
2026-06-11 16:00:39 +01:00
1857323b57
Frame the eval unit as an Agent (Model + Harness) in both READMEs
Jesse Zhang and Claude Fable 5
2026-06-11 15:22:17 +01:00
debe4105a8
Register Composer 2.5 × Grok Build TUI; fix its MythiCraft enter-world crash
Jesse Zhang and Claude Fable 5
2026-06-11 15:04:13 +01:00
c335473d93
Strip MiMo's embedded model name from its artifacts (blind-Arena identity leak)
Jesse Zhang and Claude Fable 5
2026-06-11 14:31:19 +01:00
48a8f54c6f
Add social badges to the READMEs
Jesse Zhang and Claude Fable 5
2026-06-11 00:08:54 +01:00
80a97bfe21
Regroup the registry tables by model, matching the site
Jesse Zhang and Claude Fable 5
2026-06-10 23:54:55 +01:00
8acf16262f
Rename Claude Code combination ids to the -cc suffix; refresh the Opus 4.8 mythos build
Jesse Zhang and Claude Fable 5
2026-06-10 23:07:51 +01:00
b42022e8ce
Keep CLAUDE.md local-only
Jesse Zhang and Claude Fable 5
2026-06-10 22:43:23 +01:00
42efab56e2
Register the Opus 4.8 mythos-craft run (Claude Code, Max)
Jesse Zhang and Claude Fable 5
2026-06-10 22:42:31 +01:00
664a7115c7
Become the data-only repo: site and worker move to private nagi-bench-site
Jesse Zhang and Claude Fable 5
2026-06-10 22:35:53 +01:00
ee081a4e34
Mirror case definitions at the repo root
Jesse Zhang and Claude Fable 5
2026-06-10 22:23:32 +01:00
5694580fe4
Confirm Opus 4.8 provenance (Claude Code, Max); fresh Turnstile token per sync batch
Jesse Zhang and Claude Fable 5
2026-06-10 20:50:24 +01:00
4e4f801f1f
Register the Opus 4.8 pelican (provenance TBD); ignore wrangler local state
Jesse Zhang and Claude Fable 5
2026-06-10 20:39:03 +01:00
bf23fc4168
One ballot per matchup + rating-aware matchmaking; register Opus 4.8 as pending
Jesse Zhang and Claude Fable 5
2026-06-10 20:38:10 +01:00
532985fbc3
Correct Kimi Code effort to Thinking; standardize worker docs on bunx
Jesse Zhang and Claude Fable 5
2026-06-10 20:13:54 +01:00
6ea3916b83
Add Arena blind battles, model-grouped Demo Show, and a community leaderboard
Jesse Zhang and Claude Fable 5
2026-06-10 19:49:36 +01:00
844ab3b63f
Split LpIuma's AI Studio Gemini runs into their own combinations; add PR template; register Kimi mythos run
Jesse Zhang and Claude Fable 5
2026-06-10 19:10:55 +01:00
b7d3badc72
Add four pelican runs: Fable 5 (Claude Code), Grok Build, Kimi K2.6, MiniMax M3
Jesse Zhang and Claude Fable 5
2026-06-10 18:24:32 +01:00
b008398b82
Register the Claude Code x Fable 5 Max combination as its own entry
Jesse Zhang and Claude Fable 5
2026-06-10 16:56:45 +01:00
4da51805b3
Rename the Codex CLI entry's model label back to GPT-5.5
Jesse Zhang and Claude Fable 5
2026-06-10 16:35:13 +01:00
52135f4084
Enlarge reading type; add brand icons and per-run metadata chips
Jesse Zhang and Claude Fable 5
2026-06-10 13:44:16 +01:00
2e03012b74
Add GPT-5.5 Pro pelican run
Jesse Zhang and Claude Fable 5
2026-06-10 13:07:50 +01:00
50ff7f945d
Contributor credits, SVG prefetch, integrate Xiaomi MiMo from PR #4
Jesse Zhang and DuJiabao
2026-06-10 13:03:01 +01:00
6e76858aad
Model x harness x effort registry; bigger top-bar controls
Jesse Zhang and Claude Fable 5
2026-06-10 12:09:09 +01:00
e87ef7e70f
Integrate community pelican runs from PRs #2 and #3 ; run variants
2026-06-10 12:09:09 +01:00
80817b8a2b
Data-only contributions: per-model JSON registry, validation, PR CI
Jesse Zhang and Claude Fable 5
2026-06-10 11:48:18 +01:00
e2f03163de
Add DeepSeek V4 Pro runs for both cases (#1 )
2026-06-10 18:41:05 +08:00
6f710ad413
Ignore local agent skills and skills lockfile
Jesse Zhang and Claude Fable 5
2026-06-10 03:09:02 +01:00
cd2af84f34
Viewer polish, cursor fix, Grok rename + W/S fix, Composer G teleport
Jesse Zhang and Claude Fable 5
2026-06-10 00:34:24 +01:00
37f4c8ae86
Fix Gemini 3.5 Flash atlas flip and GPT-5.5 xhigh movement mirror
Jesse Zhang and Claude Fable 5
2026-06-10 00:20:53 +01:00
54896103d8
Add GPT-5.5 Pro/xhigh and Gemini 3.5 Flash runs; fix three artifacts
Jesse Zhang and Claude Fable 5
2026-06-10 00:12:24 +01:00
f0869f0670
Add Gemini 3.1 Pro, Grok and Composer 2.5 runs for Mythos Craft
Jesse Zhang and Claude Fable 5
2026-06-09 23:50:48 +01:00
1e41d35e88
Move to bench.nagi.fun with simplified URL scheme
Jesse Zhang and Claude Fable 5
2026-06-09 22:53:35 +01:00
642bceb493
Serve from custom domain bench.nagi.fun (base: /)
Jesse Zhang and Claude Fable 5
2026-06-09 22:46:10 +01:00
06ba93a94b
Initial release: NAGI BENCH eval showcase
Jesse Zhang and Claude Fable 5
2026-06-09 22:20:15 +01:00
