GPT-6.1 Sol vs Claude Opus 5.5: API cost
Short answer: For a 10,000-input / 1,000-output request, GPT-6.1 Sol is 2× cheaper than Claude Opus 5.5. Per 1M tokens, GPT-6.1 Sol charges $2 in / $10 out; Claude Opus 5.5 charges $4 in / $20 out (checked 2026-11-18). A 2,000-input / 500-output chat call costs $0.0090 on GPT-6.1 Sol and $0.0180 on Claude Opus 5.5.
Nearest-priced models, this workload
Showing the 9 models closest in price; change any input to rank every model.
| Model | Per call | Per 1K calls |
|---|---|---|
| DeepSeek V4 Pro | $0.00462 | $4.62 |
| Grok 4.7 | $0.0070 | $7.00 |
| Gemini 2.5 Pro | $0.0075 | $7.50 |
| GPT-4.1 (deprecated) | $0.0080 | $8.00 |
| GPT-6.1 Sol | $0.0090 | $9.00 |
| GPT-6 Sol | $0.0090 | $9.00 |
| Claude Sonnet 5.5 | $0.0090 | $9.00 |
| Claude Sonnet 5 | $0.0090 | $9.00 |
| Gemini 4 Carbon | $0.0090 | $9.00 |
o4-mini is shut down; its last published list rate is kept for comparing old invoices, not as a model you can call today — the model page names its replacement.
Side by side
| GPT-6.1 Sol | Claude Opus 5.5 | |
|---|---|---|
| Input /1M | $2 | $4 |
| Cached /1M | $0.1 | $0.2 |
| Output /1M | $10 | $20 |
| Context window | 1.05M | 1M |
| Max output | 128K | 128K |
| Knowledge cutoff | Apr 30, 2026 | Jun 2026 |
| Batch input / output | $1 / $5 | $2 / $10 |
| Long-context tier | >272K: $4 / $15 | none listed |
| Status | current | Latest |
Cost for 1,000 calls
| Workload | GPT-6.1 Sol | Claude Opus 5.5 |
|---|---|---|
| Chat · 2K/500 ×1,000 | $9.00 | $18.00 |
| RAG · 20K/1K ×1,000 | $50.00 | $100 |
| RAG, 80% cached ×1,000 | $19.60 | $39.20 |
| RAG via Batch API ×1,000 | $25.00 | $50.00 |
| Generation · 500/4K ×1,000 | $41.00 | $82.00 |
| Long prompt · 300K/4K ×1,000 | $1,260 | $1,280 |
Which costs less, and when
- For a 10K-in / 1K-out request, GPT-6.1 Sol is 2× cheaper than Claude Opus 5.5.
- For output-heavy work (500 in / 4K out), GPT-6.1 Sol is cheaper; output costs 5× input on GPT-6.1 Sol and 5× on Claude Opus 5.5.
- GPT-6.1 Sol switches the whole request to its long-context rate when the prompt is over 272K tokens.
How each provider positions it
- GPT-6.1 Sol — OpenAI: “Near-Astra performance for complex work at a lower cost.”
- Claude Opus 5.5 — Anthropic: “For long-running agentic coding and knowledge work.”
Head-to-head: the two newest premium models
GPT-6.1 Sol and Claude Opus 5.5 are the models OpenAI and Anthropic currently position for hard, multi-step work. OpenAI describes GPT-6.1 Sol as “near-Astra performance for complex work at a lower cost”; Anthropic describes Opus 5.5 (released September 22, 2026) as built “for long-running agentic coding and knowledge work”. Both list a 128K max output and about a 1M-token context (1.05M for GPT-6.1 Sol, with 922K max input; 1M for Opus 5.5).
Price per token
Opus 5.5 lists exactly 2× GPT-6.1 Sol on standard input ($4 vs $2 per 1M) and output ($20 vs $10). Cache reads are $0.20 vs $0.10. Both providers give 50% off through their Batch APIs.
Where the gap narrows or widens
- Prompts over 272K tokens. OpenAI bills the whole GPT-6.1 Sol request at $4 in / $15 out above 272K input tokens. Anthropic's pricing page lists no long-context tier for Opus 5.5, so for very long prompts the input price is the same $4 on both and the output gap shrinks from 2× to 1.33×.
- Cache writes. GPT-6.1 Sol charges $2.50 per 1M to write a 30-minute cache. Opus 5.5 charges $5 for a 5-minute cache or $8 for a 1-hour cache.
- Very long batch outputs. Anthropic lets Opus 5.5 produce up to 300K output tokens per request on the Message Batches API with the
output-300k-2026-03-24beta header. OpenAI lists 128K max output for GPT-6.1 Sol. - Knowledge cutoff. OpenAI lists April 30, 2026 for GPT-6.1 Sol; Anthropic lists June 2026 for Opus 5.5.
Worked example: a 40-turn coding-agent session
Assume a 60K-token prefix (repo map, tools, instructions) is written to the cache on turn 1 and read from it on every later turn. Each later turn adds 3K fresh input tokens, and every turn returns 2K output tokens. Using the official rates, one session costs $1.42 on GPT-6.1 Sol and $2.84 on Opus 5.5. That is $1,418 vs $2,836 per 1,000 sessions. Output is 56% of the GPT-6.1 Sol bill, so capping answer length saves more than trimming the prompt.
Bottom line on cost: GPT-6.1 Sol costs about half as much for typical prompts under 272K tokens. Opus 5.5 is worth the premium only if it finishes your agent tasks in noticeably fewer turns or with fewer retries. Run both on the same 20–50 real tasks and compare cost per completed task, not cost per token.
Price is one factor; test quality on your own prompts. Sources: OpenAI pricing, Anthropic pricing, GPT-6.1 Sol docs, Claude Opus 5.5 docs (checked 2026-11-18).
GPT-6.1 Sol: ≤272K input; long context $4/$15
Claude Opus 5.5: Cache read 0.05x input
Related
Questions
Which is cheaper, GPT-6.1 Sol or Claude Opus 5.5?
For 10,000 input and 1,000 output tokens, GPT-6.1 Sol costs $0.0300 and Claude Opus 5.5 costs $0.0600, so GPT-6.1 Sol is about 2× cheaper. For output-heavy work (500 in / 4K out), GPT-6.1 Sol is cheaper.
How much do 1,000 calls cost on each?
At 2,000 input / 500 output tokens per call, 1,000 calls cost $9.00 on GPT-6.1 Sol and $18.00 on Claude Opus 5.5. At 20,000 input / 1,000 output, they cost $50.00 and $100.
How do cached input prices compare?
GPT-6.1 Sol: $0.1 per 1M cached input tokens. Claude Opus 5.5: $0.2 per 1M cached input tokens.
What does a long-context request cost?
A 100,000-input / 1,000-output request costs $0.2100 on GPT-6.1 Sol and $0.4200 on Claude Opus 5.5 at standard rates. GPT-6.1 Sol lists a 1.05M context window; Claude Opus 5.5 lists 1M.
Is there a batch discount, and what should I check before switching?
Batch pricing is published for GPT-6.1 Sol, Claude Opus 5.5 in our dataset; the provider applies it only to asynchronous batch jobs. Before switching, run your own prompts on both, check the output-token share (output usually dominates the bill) and re-check the prices on the date shown at the bottom of this page.
Which should I pick for a coding agent, GPT-6.1 Sol or Claude Opus 5.5?
On cost alone, a 60,000-token agent context with a 4,000-token answer costs $0.1600 on GPT-6.1 Sol and $0.3200 on Claude Opus 5.5 at list rates. Quality decides the rest: test both on the same real tasks and compare cost per completed task, because a model that needs a retry is never the cheaper one.
Next step → Turn this into a monthly budget
Price data last checked 2026-11-18 against official sources · how we verify
Price-change alerts
One short email when a provider changes a price or a promo ends. No spam, unsubscribe any time.