Favur Evals
favur.dev

Statement of Work

circles

59 runs, every one scored by the same deterministic engine under rubric v15.Open the interactive leaderboard for circles →

Runs against circles, ranked by composite score
#RunDateCompositeCoverage
1Gemini Flash 3.5 + Gemini Flash LiteJul 202681.8
2Mimo v2.5Jul 202681.2
3Muse Spark 1.1Jul 202680.8
4Plus qwen3.7Jul 202679.9
5Gemini Flash 3.5 + Gemini Flash LiteJul 202679.6
6Gemini Flash Lite 3.5Jul 202678.2
7Plus qwen3.7Jul 202676.7
8Gemini Flash 3.5 + Gemini Flash LiteJul 202674.6
9Gemini Flash 3.6Jul 202673.2
10Gemini Flash 3.6 + Gemini Flash LiteJul 202672.1
11Claude Sonnet 5Jul 202670.8
12Deepseek Flash v4Jul 202670.1
13Gemini Flash Lite 3.1Jul 202665.7
14Gemini Flash 3.5 + Gemini Flash LiteJul 202665.6
15Mimo v2.5Jul 202664.9
16Grok 4.5Jul 202664.7
17Mimo Pro v2.5Jul 202664.3
18Plus qwen3.7Jul 202664.3
19Mimo Pro v2.5 + MimoJul 202664.3
20Gemini Flash Lite 3.1Jul 202664.2
21Deepseek Flash v4Jul 202664.0
22Kimi k2.6Jul 202663.3
23Grok 4.5Jul 202663.2
24Mimo Pro v2.5Jun 202663.0
25Grok 4.5Jul 202662.3
26Glm 5.2Jul 202661.6
27Gpt Terra 5.6Jul 202661.5
28Kimi Code k2.7Jul 202661.1
29Gemini Flash 3.6Jul 202661.1
30Gpt Luna 5.6Jul 202661.1
31Grok 4.5Jul 202661.0
32Kimi Code k2.7Jul 202659.3
33Step Flash 3.7Jul 202658.9
34Claude Sonnet 5Jul 202657.9
35Deepseek Flash v4Jul 202656.8
36Gemini Flash 3 (preview)Jun 202656.6
37Gemini Flash 3 (preview)Jul 202656.5
38Deepseek Flash v4Jul 202656.0
39Deepseek Pro v4Jul 202655.9
40Glm 5.2Jul 202654.5
41Gemini Flash 3 (preview)Jul 202654.4
42Deepseek Pro v4Jun 202653.9
43Gemini Flash 3 (preview)Jul 202653.8
44Gemini Pro 3.1 (preview)Jul 202653.4
45Kimi k2.6Jun 202653.1
46Gemini Flash 3.5Jul 202652.6
47Deepseek Pro v4Jul 202651.9
48Gemini Pro 3.1 (preview)Jun 202651.7
49InklingJul 202651.7
50Gemini Flash Lite 3.1Jul 202651.0
51Kimi k2.6Jul 202649.9
52InklingJul 202648.6
53Claude Sonnet 5Jul 202647.8
54Gemini Flash 3.5Jul 202647.4
55Glm 5.2Jun 202646.9
56Claude Sonnet 5Jul 202646.6
57Claude Sonnet 4.6Jul 202644.4
58Claude Sonnet 4.6Jun 202642.3
59Gemini Flash 3 (preview)Jul 202639.1

Every run on Favur Evals is scored by the same deterministic engine, on the same Statements of Work. The benchmark is self-funded — no vendor sponsorship, credits, or grants.