Terminal-Bench 4.0 Results Across LLMs

70.6% · raw 70.6 %

Headline · max effort · Own vendor

Source/record date: 2026-09-28

TB 4.0 at max effort; safeguards on (1.2% of requests fallback-served); 5 trials per task

66.4% · raw 66.4 %

Headline · xhigh effort · Own vendor

Source/record date: 2026-09-22

Terminal-Bench 4.0 at xhigh (table footnote); adaptive thinking

66.4% · raw 66.4 %

Alternative · unknown effort · Peer vendor

Source/record date: 2026-09-30

As reported by Google Gemini 4 Argon chart; peer harness notes in Google eval methodology | Same value as current headline; kept as corroboration, non-headline.

57.9% · raw 57.9 %

Headline · unknown effort · Own vendor

Source/record date: 2026-09-03

57.9% · raw 57.9 %

Alternative · unknown effort · Peer vendor

Source/record date: 2026-09-22

As reported by Anthropic Opus 5.5 announce (OpenAI figure) | Demoted 2026-09-25: duplicate of headline from OpenAI GPT-6 Astra announce (openai.com/index/gpt-6-astra); vendor official preferred over Anthropic peer comparison table (rule a); same value.

58.2% · raw 58.2 %

Alternative · unknown effort · Peer vendor

Source/record date: 2026-09-30

As reported by Google Gemini 4 Argon chart; peer harness notes in Google eval methodology | Differs from current headline 57.9 (first-party); kept with provenance, non-headline.

57.4% · raw 57.4 %

Headline · max effort · Own vendor

Source/record date: 2026-09-30

Self-computed; peers from official public leaderboard (highest thinking level) | Highest thinking settings per Google eval methodology.

55.8% · raw 55.8 %

Alternative · unknown effort · Peer vendor

Source/record date: 2026-09-03

As reported by OpenAI GPT-6 Astra announcement table | Demoted 2026-09-25: duplicate of headline from Anthropic Claude Opus 5.5 announce (anthropic.com/claude-opus-5-5); model vendor's own (Anthropic) figure preferred over peer-vendor OpenAI comparison table (rule a); Anthropic source is also later-dated (2026-09-22); same value. source_type relabeled official→vendor-comparison (OpenAI reporting an Anthropic model).

55.8% · raw 55.8 %

Headline · unknown effort · Peer vendor

Source/record date: 2026-09-22

As reported by Anthropic Opus 5.5 announce

57.9% · raw 57.9 %

Alternative · unknown effort · Peer vendor

Source/record date: 2026-09-30

As reported by Google Gemini 4 Argon chart; peer harness notes in Google eval methodology | Differs from current headline 55.8 (spillover); kept with provenance, non-headline.

50.8% · raw 50.8 %

Headline · unknown effort · Own vendor

Source/record date: 2026-10-01

Unbiased preliminary vendor result; mean task cost $0.48. Unbiased says results may change and denominator and cost methods need confirmation.

40.4% · raw 40.4 %

Headline · unknown effort · Third-party

Source/record date: 2026-09-30

Via ThreatFrontier transcription of Ant launch table; original pixels not yet re-read. Effort undisclosed.

38% · raw 38 %

Headline · unknown effort · Own vendor

Source/record date: 2026-09-21

37.9% · raw 37.9 %

Headline · unknown effort · Peer vendor

Source/record date: 2026-09-10

As reported in DeepSeek-V4.1-Flash HF comparison table (GLM-5.3 column)

34.9% · raw 34.9 %

Headline · unknown effort · Own vendor

Source/record date: 2026-09-21

32.8% · raw 32.8 %

Headline · unknown effort · Third-party

Source/record date: 2026-09-30

Via ThreatFrontier transcription of Ant launch table; original pixels not yet re-read. Effort undisclosed.

31.2% · raw 31.2 %

Headline · unknown effort · Third-party

Source/record date: 2026-09-10

Secondary guide citing DeepSeek official agent snapshot

28.8% · raw 28.8 %

Headline · unknown effort · Own vendor

Source/record date: 2026-09-21

21.5% · raw 21.5 %

Headline · max effort · Own vendor

Source/record date: 2026-09-23

Official Terminal-Bench 4.0 board: GPT-5.6 Terra (max) + Codex; resolution rate 21.5% ±3.3

19.1% · raw 19.1 %

Headline · unknown effort · Own vendor

Source/record date: 2026-09-03