Log inSign up
Benchmarks › OpenAI › GPT
OpenAIclosed weightsCurrentReasoning

GPT-6.1 Terra

Mid-tier GPT-6.1 model (between Luna and Sol) for computer use, visual reasoning and writing; succeeds GPT-5.6 Terra. $1/$5 up to 272K input ($2/$7.50 above), 128K max output, cached input $0.05. OpenAI reports Terminal-Bench 4.0 50.3% (max), OSWorld 2.1 74.2%.

Released
Oct 26, 2026
Price in / out
$1 / $5
Blended $/1M
$2
Context
1.05M

Benchmark scoresvs GPT-6.1 Sol

BenchmarkScoreRank (current)Source
Humanity's Last Exam
46.2%#14 of 36artificialanalysis.ai ↗
Artificial Analysis Intelligence Index
51#5 of 33artificialanalysis.ai ↗

GPT versions

Timeline
Version$ in / outTerminal-BenchSWE-benchGPQAHLEARC-AGI-2τ²-bench
GPT-4oRetired$2.5 / $10—————25.1%
o1Retired$15 / $60—————62.6%
o3-miniRetired$1.1 / $4.4—49.3%———28.7%
GPT-4.1Superseded$2 / $8—54.6%———47.1%
o3Superseded$2 / $8—69.1%———80.7%
GPT-5Superseded$1.25 / $10—74.9%85.7%——84.8%
GPT-5.1Superseded$1.25 / $10—————81.9%
GPT-5.2Superseded$1.75 / $14—80%——52.9%84.8%
GPT-5.3 CodexSuperseded$1.75 / $14—85%———86%
GPT-5.4Superseded$2.5 / $15——92.8%39.8%73.3%98.9%
GPT-5.5Superseded$5 / $3076.4%—93.6%41.4%85%98%
GPT-5.5 ProSuperseded$30 / $180———43.1%84.2%—
GPT-5.6 LunaSuperseded$0.2 / $1.279%—92.3%—59.5%—
GPT-5.6 SolSuperseded$5 / $3085.8%—94.6%—92.5%85.1%
GPT-5.6 TerraSuperseded$2 / $1277.5%—92.9%—83.9%86.3%
GPT-6 Astra$10 / $5087.3%—96%54.7%95%—
GPT-6 Luna$0.1 / $0.573%——38.5%59.3%—
GPT-6 SolSuperseded$2 / $1083.1%——47.9%89.6%—
GPT-6.1 Sol$2 / $10———52.9%94.2%—
GPT-6.1 Terra$1 / $5———46.2%——