none reviewed yet · 2 benchmark observations · latest capture Sep 8, 2026
Inspect the evidenceOpenAI · MODEL PROFILE
GPT-5.6 Sol
Teams standardised on Codex who want frontier agentic coding without Astra pricing
Saved editorial fit for your current task lens; not a benchmark result
How scoring worksVerified rates (Sep 8, 2026) · 100 runs / month at your token settings
Adjust workloadAt a glance
9 facts- Provider
- OpenAI
- Type
- API model
- Access
- OpenAI API, Codex harness
- Context
- 1,050,000 tokens
- Maximum output
- 128,000 tokens
- License / weights
- Proprietary
- Input modalities
- file, image, text
- Output modalities
- text
- Pricing notes
- $4 in / $20 out per 1M (promotional, was $5/$30); cached input $0.40; standard rates apply below 272K context
The practical trade-offs
Documented strengths
- Highest-vote frontier Arena row in the OpenAI family (8,727 votes, +/-8)
- Terminal Bench 2.1 88.8% is the best number in Z.ai's cross-vendor table
- Solid agentic evidence: Agents' Last Exam 53.6%, OSWorld 2.0 65.7%
- Discounted while the promotion lasts ($4/$20)
Limitations to check
- Terminal-Bench 4.0 37.3% is far behind GPT-6 Astra's 57.9%
- High TTFT (107 s at max effort)
- Frontend score is Codex-harness-assisted, so model-only quality is unclear
- Promotional price expires after Nov 21, 2026
Follow the source
6 source linksEvidence & conditions
6Source links
NoneReviewed records
2Benchmark runs
Sep 8, 2026Latest capture
- Evidence source 1
developers.openai.comDate not recorded - Evidence source 2
openai.comDate not recorded - Evidence source 3
arena.aiDate not recorded - Evidence source 4
benchlm.aiDate not recorded - Evidence source 5
artificialanalysis.aiDate not recorded - Evidence source 6
openai.comDate not recorded
Benchmark observations
Terminal-Bench 4.0 Inspect conditions
- name
- Terminal-Bench 4.0
- value
- 37.3
- unit
- %
- source Url
- https://openai.com/index/gpt-6-astra/
- checked At
- 2026-09-08T19:30:56Z
- attribution
- provider-reported
- configuration
- OpenAI release table; maximum across effort; research/API evaluation, not production ChatGPT.
DeepSWE v1.1 Inspect conditions
- name
- DeepSWE v1.1
- value
- 72.7
- unit
- %
- source Url
- https://openai.com/index/gpt-6-astra/
- checked At
- 2026-09-08T19:30:56Z
- attribution
- provider-reported
- configuration
- OpenAI release table; maximum across effort; research/API evaluation, not production ChatGPT.
Source reports, measured results, and editorial assessments are separate evidence types. Source dates do not imply a new benchmark run. Capture dates are shown when recorded; otherwise we say the capture date was not recorded. Recorded dates are compared with this catalogue snapshot (Sep 11, 2026), not today’s date.
Method AIS-V2
How this page is assessed
Quality claims follow Method AIS-V2.
Money never buys rank — commercial relationships cannot change rank, eligibility, evidence quality, or publication timing.
Submit or claim a listing is free and does not skip the editorial queue.
Private · this device only