🤖 AI benchmark: hit-rate of 7 models

Prematch Live (in-play)

Seven external AI models (Hermes contour) independently analyze the same matches — predicting the outcome (1X2), total (Over/Under), both teams to score (BTTS) and the exact score. Here we honestly compare their predictions against the real result after the final whistle and combine everything into a single accuracy rating. An informational and analytical snapshot, not betting advice.

⚠️ Data is still accumulating — counting starts from 09.07.2026, so all models are compared on the same events (early test predictions are excluded). The sample is still small and not representative. Right now the snapshot holds 1032 match(es), 3693 settled AI predictions (Tennis). The figures below are N, not «a percentage you can trust»: the more matches are played out, the more reliable the snapshot becomes. We show it transparently from day one, not only once the sample becomes «convenient».

All sports Football Tennis Basketball Volleyball Hockey Table tennis

Leaderboard · Tennis

Model	N (settled)	1X2	Exact score (by sets)	Composite accuracy
ChatGPT	106	71.7%(76/106)	33.3%(28/84)	54.7%(104/190)
Kimi	508	66.9%(340/508)	33.9%(151/445)	51.5%(491/953)
DeepSeek	84	69.0%(58/84)	32.9%(27/82)	51.2%(85/166)
Claude	760	65.8%(500/760)	35.2%(248/704)	51.1%(748/1464)
Google AI	1060	64.8%(687/1060)	35.7%(374/1049)	50.3%(1061/2109)
GLM 5.2	573	63.5%(364/573)	35.1%(196/559)	49.5%(560/1132)
Qwen	602	62.5%(376/602)	28.2%(163/579)	45.6%(539/1181)

grey — sample <5, not representative; «—» — the model has not made a settled prediction yet.

Composite accuracy — the share of correct predictions across all shown markets together: (sum of correct picks) ÷ (sum of all settled picks) across the markets 1X2 + Exact score (by sets). Each market-pick weighs equally. This is hit-rate, not profitability — for money/ROI by model see /ai-agent. «Exact score (by sets)» — the full final score was guessed correctly (H and A matched) by sets; predictions with no recognized score do not count toward the denominator.

Composite model rating · all markets · Tennis

Bar height = the model's composite accuracy across all applicable markets on the current sample. Sorted from best to worst.

54.7% (104/190)

GPT 5.5

51.5% (491/953)

Kimi 2.6

51.2% (85/166)

DeepSeek V4 Pro

51.1% (748/1464)

Opus 4.8

50.3% (1061/2109)

Gemini 3.5 Flash

49.5% (560/1132)

GLM 5.2

45.6% (539/1181)

Qwen 3.7 Plus

Bars are AI models by version; grey/dimmed — sample <5, not representative. The snapshot is informational, not betting advice.

Accuracy by market · Tennis

Where each model is strong: one mini-bar per applicable market, with the percentage and (hits/sample).

ChatGPT Composite 54.7%

1X2

71.7% (76/106)

Exact score (by sets)

33.3% (28/84)

Kimi Composite 51.5%

1X2

66.9% (340/508)

Exact score (by sets)

33.9% (151/445)

DeepSeek Composite 51.2%

1X2

69.0% (58/84)

Exact score (by sets)

32.9% (27/82)

Claude Composite 51.1%

1X2

65.8% (500/760)

Exact score (by sets)

35.2% (248/704)

Google AI Composite 50.3%

1X2

64.8% (687/1060)

Exact score (by sets)

35.7% (374/1049)

GLM 5.2 Composite 49.5%

1X2

63.5% (364/573)

Exact score (by sets)

35.1% (196/559)

Qwen Composite 45.6%

1X2

62.5% (376/602)

Exact score (by sets)

28.2% (163/579)

The model's favorite by 1X2 = the max of P1/X/P2 in its probabilities; for sports without a draw (tennis, volleyball, etc.) the «X» option doesn't participate. grey — sample <5, not representative. Not betting advice.