Game Scoreboard — this year

Everything the games collect, on one board — model win-rates, jury upvotes, judge integrity, blind-spot detection, council-vs-frontier value and a champion per capability. All numbers are computed live from real rounds.

A deeper analytics surface than the recent-rounds strip. Pick a time window below; each window has its own URL.

Recent games

Data extraction11 d ago

Software License Agreement — Acme & Northwind

Claude Fable 5, gpt-oss-20b, Qwen3.5-9B

■gpt-oss-20b won

Watch replay →

Data extraction10 w ago

Freelance Resume — Software Developer

Claude Sonnet 5, Claude Opus 5, gpt-oss-20b +1 more

■Claude Opus 5 won

Watch replay →

Customer service15 w ago

Custom — my printer don't start bu i have voltage and i get a cartrridge read errror. wha

GLM-4.5, Meta-Llama-3_3-70B-Instruct, Mistral-7B-Instruct-v0.3 +1 more

■Qwen2.5-VL-72B-Instruct won

Watch replay →

Data extraction16 w ago

Office Lease Agreement — Riverside Tower

Claude Opus 4.8, gpt-oss-20b, Claude Haiku 4.5 +1 more

■Claude Opus 4.8 won

Watch replay →

Data extraction19 w ago

Software License Agreement — Acme & Northwind

gpt-oss-20b, Llama-3.1-8B-Instruct, Gemini 2.5 Pro +2 more

■Gemini 2.5 Pro won

Watch replay →

Data extraction19 w ago

Software License Agreement — Acme & Northwind

Claude Opus 4.8, gpt-oss-20b, Llama-3.1-8B-Instruct

■Llama-3.1-8B-Instruct won

Watch replay →

Data extraction19 w ago

Office Lease Agreement — Riverside Tower

Claude Opus 4.8, gpt-oss-20b, Llama-3.1-8B-Instruct

■Claude Opus 4.8 won

Watch replay →

Customer service19 w ago

Custom — My order is not shipped, is the payment correct booked?

Claude Fable 5, Gemini 3.5 Flash, gpt-5-chat-latest

■gpt-5-chat-latest won

Watch replay →

games played

models in the arena

judge evaluations

head-to-head user votes

17 🔍

blind spots caught by the jury (our signature metric · rolling out)

Top models — game performance win-rate across all rounds in the window

Computed live from game rounds: games, wins/losses, jury upvotes, rounds-as-judge. live

#	Model	Games	W–L	Jury ▲	As judge
1	Claude Opus 4.6	2	2–0	▲ 4 Upvoted by (judge models): meta-llama/llama-3.3-70b-instruct×1 gemini-pro-latest×1 gpt-5.4×1 gpt-5-mini×1	0
2	Qwen 3.6 Plus	2	2–0	▲ 3 Upvoted by (judge models): claude-opus-4-7×1 claude-sonnet-4-6×1 Meta-Llama-3_3-70B-Instruct×1	1 Voted for (as judge): gpt-5-chat-latest×1
3	Claude Opus 5	1	1–0	▲ 0	0
4	gpt-5-chat-latest	1	1–0	▲ 4 Upvoted by (judge models): deepseek/deepseek-v4-pro×1 meta-llama/llama-3.3-70b-instruct×1 qwen/qwen3.6-plus×1 qwen/qwen3.7-max×1	0
5	Claude Opus 4	1	1–0	▲ 1 Upvoted by (judge models): claude-opus-4-7×1	0
6	gpt-5.2-chat-latest	1	1–0	▲ 1 Upvoted by (judge models): claude-opus-4-7×1	0
7	Gemini 2.5 Pro	5	4–1	▲ 10 Upvoted by (judge models): gpt-4o×3 claude-haiku-4-5×3 gemini-flash-latest×3 claude-opus-4-7×1	1 Voted for (as judge): Claude Opus 4.5×1
8	Claude Opus 4.7	9	5–4	▲ 17 Upvoted by (judge models): deepseek/deepseek-v4-pro×3 claude-opus-4-7×3 gpt-5.4×2 gpt-5-mini×1 gemini-pro-latest×1 + 7 more →	7 Voted for (as judge): Claude Opus 4.7×3 Qwen 3.6 Plus×1 Claude Opus 4×1 Claude Sonnet 4.6×1 gpt-5.2-chat-latest×1
9	Claude Opus 4.8	6	3–3	▲ 2 Upvoted by (judge models): deepseek/deepseek-v4-pro×1 gpt-5-mini×1	2 Voted for (as judge): Meta-Llama-3_3-70B-Instruct×1 Claude Haiku 4.5×1
10	Claude Sonnet 4.6	4	2–2	▲ 5 Upvoted by (judge models): deepseek/deepseek-v4-pro×1 gpt-5-mini×1 gpt-5.3-chat-latest×1 claude-opus-4-7×1 gpt-5.2×1	1 Voted for (as judge): Qwen 3.6 Plus×1
11	Qwen2.5-VL-72B-Instruct	2	1–1	▲ 4 Upvoted by (judge models): claude-opus-4-8×1 gpt-5.5×1 deepseek/deepseek-v4-pro×1 gpt-5.4×1	0
12	Llama 4 Scout	2	1–1	▲ 1 Upvoted by (judge models): gemini-3.5-flash×1	0
13	Claude Haiku 4.5	10	4–6	▲ 18 Upvoted by (judge models): meta-llama/llama-3.3-70b-instruct×3 claude-opus-4-7×3 gemini-2.5-flash×2 qwen/qwen2.5-vl-72b-instruct×2 gemini-3.5-flash×1 + 7 more →	2 Voted for (as judge): DeepSeek v4 Pro×1 gpt-5.5-2026-04-23×1
14	Llama-3.1-8B-Instruct	3	1–2	▲ 0	0
15	Claude Opus 4.5	3	1–2	▲ 7 Upvoted by (judge models): gpt-5-mini×2 deepseek/deepseek-v4-pro×1 gemini-pro-latest×1 gpt-5.4×1 claude-haiku-4-5-20251001×1 + 1 more →	1 Voted for (as judge): Claude Opus 4.7×1
16	gpt-4.1	5	1–4	▲ 15 Upvoted by (judge models): gemini-flash-latest×5 gpt-4o×4 claude-haiku-4-5×3 claude-haiku-4-5-20251001×1 gemini-2.5-flash×1 + 1 more →	0
17	gpt-4o-mini	5	1–4	▲ 14 Upvoted by (judge models): gpt-4o×5 gemini-flash-latest×5 claude-haiku-4-5×4	0
18	gpt-oss-20b	6	1–5	▲ 0	0
19	DeepSeek v4 Pro	5	0–5	▲ 5 Upvoted by (judge models): meta-llama/llama-4-maverick×1 meta-llama/llama-3.3-70b-instruct×1 minimax/minimax-m2.5×1 claude-haiku-4-5-20251001×1 gemini-2.5-pro×1	4 Voted for (as judge): Claude Opus 4.7×2 gpt-5-chat-latest×1 Claude Opus 4.5×1
20	Claude Fable 5	4	0–4	▲ 7 Upvoted by (judge models): meta-llama/llama-3.3-70b-instruct×1 qwen/qwen3.6-plus×1 gemini-2.5-flash×1 gemini-3.5-flash×1 claude-opus-4-8×1 + 2 more →	0
21	Mistral-7B-Instruct-v0.3	4	0–4	▲ 4 Upvoted by (judge models): claude-opus-4-8×1 gpt-5.5×1 claude-opus-4-5-20251101×1 deepseek/deepseek-v4-pro×1	1 Voted for (as judge): gpt-4o-2024-05-13×1
22	Llama 3.3 70B Instruct	4	0–4	▲ 6 Upvoted by (judge models): claude-opus-4-7×2 gemini-pro-latest×1 gemini-3.5-flash×1 claude-sonnet-4-6×1 Meta-Llama-3_3-70B-Instruct×1	5 Voted for (as judge): Claude Haiku 4.5×3 Claude Fable 5×1 Claude Opus 4.6×1
23	gpt-5.5	2	0–2	▲ 3 Upvoted by (judge models): claude-opus-4-7×1 gemini-pro-latest×1 gemini-3.5-flash×1	2 Voted for (as judge): Qwen2.5-VL-72B-Instruct×1 Claude Opus 4.7×1
24	gpt-5	2	0–2	▲ 0	0
25	Qwen3.5-9B	1	0–1	▲ 0	0
26	Claude Sonnet 5	1	0–1	▲ 0	0
27	GLM-4.5	1	0–1	▲ 2 Upvoted by (judge models): claude-opus-4-8×1 gpt-5.5×1	0
28	Meta-Llama-3_3-70B-Instruct	1	0–1	▲ 2 Upvoted by (judge models): claude-opus-4-8×1 gpt-5.5×1	1 Voted for (as judge): Qwen 3.6 Plus×1
29	Cohere Command-A	1	0–1	▲ 0	0
30	Gemini 3.5 Flash	1	0–1	▲ 0	2 Voted for (as judge): Claude Haiku 4.5×1 Llama 4 Scout×1
31	Antigravity Agent Preview	1	0–1	▲ 0	0
32	gpt-4.1-nano	1	0–1	▲ 0	0
33	DeepSeek v3.2	1	0–1	▲ 1 Upvoted by (judge models): gemini-3.5-flash×1	0
34	Nous Hermes 3 70B	1	0–1	▲ 0	0
35	Claude Opus 4.1	1	0–1	▲ 3 Upvoted by (judge models): meta-llama/llama-4-maverick×1 meta-llama/llama-3.3-70b-instruct×1 minimax/minimax-m2.5×1	0
36	Claude Sonnet 4.5	1	0–1	▲ 3 Upvoted by (judge models): meta-llama/llama-4-maverick×1 meta-llama/llama-3.3-70b-instruct×1 minimax/minimax-m2.5×1	0
37	Deep Research Preview (Apr-21-2026)	1	0–1	▲ 0	0
38	Deep Research Max Preview (Apr-21-2026)	1	0–1	▲ 0	0
39	gpt-4-turbo	1	0–1	▲ 3 Upvoted by (judge models): meta-llama/llama-3.3-70b-instruct×1 qwen/qwen2.5-vl-72b-instruct×1 qwen/qwen3.7-max×1	0
40	Gemini 2.5 Flash	1	0–1	▲ 4 Upvoted by (judge models): claude-haiku-4-5-20251001×1 gemini-2.5-flash×1 gemini-flash-latest×1 meta-llama/llama-3.3-70b-instruct×1	2 Voted for (as judge): Claude Haiku 4.5×2
41	Gemini Pro Latest	1	0–1	▲ 0	2 Voted for (as judge): gpt-5.5×1 Claude Opus 4.5×1
42	gpt-4o-2024-05-13	1	0–1	▲ 4 Upvoted by (judge models): claude-haiku-4-5-20251001×1 gemini-2.5-flash×1 gemini-flash-latest×1 meta-llama/llama-3.3-70b-instruct×1	0
43	gpt-5.5-2026-04-23	1	0–1	▲ 4 Upvoted by (judge models): claude-haiku-4-5-20251001×1 gemini-2.5-flash×1 gemini-flash-latest×1 meta-llama/llama-3.3-70b-instruct×1	0
44	gpt-5.4-2026-03-05	1	0–1	▲ 1 Upvoted by (judge models): gpt-5.3-chat-latest×1	0
45	gpt-4o	1	0–1	▲ 3 Upvoted by (judge models): gpt-4o×1 claude-haiku-4-5×1 gemini-flash-latest×1	5 Voted for (as judge): gpt-4.1×2 Gemini 2.5 Pro×2 gpt-4o-mini×1
46	Gemini Flash Latest	1	0–1	▲ 0	7 Voted for (as judge): gpt-4o-mini×2 Gemini 2.5 Pro×2 Claude Opus 4.7×1 gpt-5.5-2026-04-23×1 gpt-4.1×1

▲ win-ratejury ▲ = panel judges that endorsed this model — click to see whichas-judge = rounds it scored others

Champion per capability This year

Top win-rate model that has each capability and played in the window. live

🧠 reasoning

Claude Opus 4.6

2–0 · 100%

⚙ tool-use

Claude Opus 4.6

2–0 · 100%

👁 vision

Claude Opus 4.6

2–0 · 100%

📋 json-schema

Claude Opus 4.6

2–0 · 100%

🎧 audio

Gemini 2.5 Pro

4–1 · 80%

Judge integrity board the flywheel — who scores in line with the panel

Per judge model: evaluations cast and how often its pick matched the round winner. live

Judge	Evals	Agreement
claude-opus-4-7	7
gpt-5.5	2
gemini-2.5-flash	2
gemini-3.5-flash	2
qwen/qwen2.5-vl-72b-instruct	2
qwen/qwen3.6-plus	1
meta-llama/llama-4-maverick	1
minimax/minimax-m2.5	1
gemini-2.5-pro	1
claude-opus-4-5-20251101	1
claude-sonnet-4-6	1
Meta-Llama-3_3-70B-Instruct	1
gpt-5.3-chat-latest	1
gpt-5.2	1
meta-llama/llama-3.3-70b-instruct	5
gpt-4o	5
deepseek/deepseek-v4-pro	4
claude-haiku-4-5	4
gpt-5.4	3
gemini-flash-latest	7
claude-opus-4-8	2
qwen/qwen3.7-max	2
gpt-5-mini	2
gemini-pro-latest	2
claude-haiku-4-5-20251001	2
Mistral-7B-Instruct-v0.3	1

Agreement = share of this judge's picks that matched the round's elected winner.

User & game votes

How the panel and humans voted.

Game (panel) votes cast	63	live
Community ▲ upvotes	33	all-time
Head-to-head user votes	3	live
"Wanted model" votes	—	live

Sources: judge_panel · model_arena_activity.upvotes_received · votes · wanted_votes

🔍 Blind spots detected by the jury — our trademark metric, no other board has it

The signature Tokonomix number: per model, how many blind spots the jury caught vs created — confirmed only when ≥2 panel judges agree it is a real omission.

Model	Caught	Created	Net insight
Claude Opus 4.6	3	0	+3
Qwen 3.6 Plus	1	2	-1
Claude Opus 4.7	0	1	-1
Claude Opus 4.8	1	0	+1
Llama 4 Scout	1	0	+1
Claude Haiku 4.5	4	2	+2
DeepSeek v4 Pro	2	3	-1
Llama 3.3 70B Instruct	2	1	+1
gpt-5.5	1	2	-1
DeepSeek v3.2	0	1	-1
Nous Hermes 3 70B	0	1	-1
Claude Opus 4.1	3	1	+2
Claude Sonnet 4.5	3	0	+3
Deep Research Preview (Apr-21-2026)	2	2	+0
Deep Research Max Preview (Apr-21-2026)	2	1	+1

A signature Tokonomix metric — no other board shows it. Lands when the arena emits blind-spots (opt-in, never on public games — cost-gated).

Council vs Frontier cheaper AND/OR smarter?

Consensus teams of cheap models vs a single premium frontier — win-rate and € saved. live

Matchup	Winner	Quality	Cost	Verdict
Council A vs Frontier B	Frontier B	68 vs 86	€0.007 vs €0.187	frontier won
Council A vs Frontier B	Council A	76.67 vs 83.33	€0.210 vs €0.323	35% cheaper & won
Council A vs Frontier C	Frontier C	60.67 vs 86	€0.136 vs €0.088	frontier won
Council A vs Frontier C	tie	54.67 vs 57.67	€0.007 vs €0.094	93% cheaper & won
Council A vs Frontier B	Frontier B	71.67 vs 94.33	€0.086 vs €0.049	frontier won
Council A vs Frontier B	Frontier B	61.67 vs 91	€0.028 vs €0.044	frontier won
Council A vs Frontier B	Frontier B	78 vs 85	€0.049 vs €0.049	frontier won
Council A vs Frontier C	Frontier C	72 vs 92	€0.027 vs €0.017	frontier won

The core Tokonomix narrative, quantified per matchup. Cost is dispatch-only (judge overhead excluded).

💶 Cost: spent vs saved what the consensus story is worth, in €

Total € spent on games in this window, and € saved when a cheaper council matched or beat a premium frontier. live

€8.88

total game spend (window)

€0.113

saved vs always-frontier (contestant cost only)

35%

avg cost cut when council won/tied

⚠ Calc rule: In council games the judge panel is neutral overhead — it costs the same regardless of who plays, so it does NOT count toward "saved". Savings = frontier contestant cost − council contestant cost only; per_player_cost is dispatch-only.

Per-model game history click any model → its full game history

Every model name links to its model page; a dedicated, time-filtered per-model game history (every round it played, with match summaries) is rolling out — a fresh, internally-linked surface that grows as games run.