GPT-5.6 Sol
Language AvailableCompare models
ComparingGPT-5.6 Solwith
Pick a model above to see the comparison.
openai/gpt-5.6-sol · by OpenAI · decoder-only-transformer
Pricing — 1 offering(s)
Input tokens
- $5.00 / 1M tokens (input) Current 2026-07-10 → present
Output tokens
- $30.00 / 1M tokens (output) Current 2026-07-10 → present
Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.
Capability profile
reasoning strong
coding strong
tool use strong
context window strong
speed weak
cost efficiency weak
Coding benchmarks
View full leaderboard →| Benchmark | Score | Harness | Source |
|---|---|---|---|
| swe-bench-verified ↗ | 96.2% 2026-07 | — | 📊 leaderboard ↗ |
Operator guidance
OpenAI's new flagship, succeeding gpt-5.5 at the same $5/$30 price point (launched 2026-07-10). Route here for top-tier reasoning and coding. Drop to Terra for a balanced cost/intelligence tradeoff, or Luna for cost-sensitive, high-volume workloads.
Use cases
- Complex reasoning, coding, and professional work — OpenAI's recommended tier for these use cases
- Frontier general-purpose and agentic tasks, same positioning as the gpt-5.5 it succeeds
Limitations
- instruction-following and multilingual ratings omitted — checked 2026-07-20 (live browser fetch of openai.com/index/gpt-5-6/ in full, the OpenAI Help Center's 'GPT-5.6 in ChatGPT' article, and the GPT-5.6 system card at deploymentsafety.openai.com/gpt-5-6) and no distinct per-tier claim exists on any OpenAI-owned page for either dimension; rather than infer a rating, this doc leaves them out
- speed rating added 2026-07-20 (previously omitted) — the GA launch page's 'Availability and pricing' section explicitly calls Luna 'our fastest' model; Sol, the flagship/highest-reasoning tier, is rated 'weak' by elimination as the slowest of the three — a within-family relative rating, not a distinct sourced claim for Sol itself
- Capability ratings are qualitative, not a single cited benchmark run
- SCO-250: OpenAI's GPT-5.6 Sol preview announcement (openai.com/index/previewing-gpt-5-6-sol/, required a live browser fetch — flat HTTP 403s) states verbatim: 'We're also launching GPT-5.6 Sol on Cerebras at up to 750 tokens per second in July, bringing frontier intelligence to customers at unprecedented speed. Access will initially be limited to select customers as we expand capacity.' This is a distinct, capacity-limited, select-customers-only Cerebras hardware deployment ('Sol Fast') — not the throughput of the standard API model this registry prices and tracks, and no equivalent figure exists for Terra or Luna. Deliberately not folded into the `speed` capability_profile rating above, which describes the standard deployment only.