DeepSeek V4 Pro vs GPT-6.1 Sol: API cost
Short answer: For a 10,000-input / 1,000-output request, DeepSeek V4 Pro is 1.7× cheaper than GPT-6.1 Sol. Per 1M tokens, DeepSeek V4 Pro charges $1.32 in / $3.96 out; GPT-6.1 Sol charges $2 in / $10 out (checked 2026-11-18). A 2,000-input / 500-output chat call costs $0.00462 on DeepSeek V4 Pro and $0.0090 on GPT-6.1 Sol.
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 |
|---|---|---|
| Grok 4.3 | $0.00375 | $3.75 |
| o4-mini (shut down 2026-10-23) | $0.0044 | $4.40 |
| GPT-6.1 Terra | $0.0045 | $4.50 |
| Claude Haiku 4.5 | $0.0045 | $4.50 |
| 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 |
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
| DeepSeek V4 Pro | GPT-6.1 Sol | |
|---|---|---|
| Input /1M | $1.32 | $2 |
| Cached /1M | $0.044 | $0.1 |
| Output /1M | $3.96 | $10 |
| Context window | 1M | 1.05M |
| Max output | 384K | 128K |
| Knowledge cutoff | – | Apr 30, 2026 |
| Batch input / output | no batch | $1 / $5 |
| Long-context tier | none listed | >272K: $4 / $15 |
| Status | current | current |
Cost for 1,000 calls
| Workload | DeepSeek V4 Pro | GPT-6.1 Sol |
|---|---|---|
| Chat · 2K/500 ×1,000 | $4.62 | $9.00 |
| RAG · 20K/1K ×1,000 | $30.36 | $50.00 |
| RAG, 80% cached ×1,000 | $9.94 | $19.60 |
| RAG via Batch API ×1,000 | no batch | $25.00 |
| Generation · 500/4K ×1,000 | $16.50 | $41.00 |
| Long prompt · 300K/4K ×1,000 | $412 | $1,260 |
Which costs less, and when
- For a 10K-in / 1K-out request, DeepSeek V4 Pro is 1.7× cheaper than GPT-6.1 Sol.
- For output-heavy work (500 in / 4K out), DeepSeek V4 Pro is cheaper; output costs 3× input on DeepSeek V4 Pro and 5× on GPT-6.1 Sol.
- 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
- DeepSeek V4 Pro — DeepSeek: “DeepSeek's higher-priced V4 Pro tier, same 1M context.”
- GPT-6.1 Sol — OpenAI: “Near-Astra performance for complex work at a lower cost.”
Price is one factor; test quality on your own prompts. Sources: DeepSeek pricing, OpenAI pricing, DeepSeek V4 Pro docs, GPT-6.1 Sol docs (checked 2026-11-18).
DeepSeek V4 Pro: Peak rates; off-peak is half (peak = 01:00-04:00 and 06:00-10:00 UTC, Mon-Fri). Served by DeepSeek-V4-Pro-0813.
GPT-6.1 Sol: ≤272K input; long context $4/$15
Related
Questions
Which is cheaper, DeepSeek V4 Pro or GPT-6.1 Sol?
For 10,000 input and 1,000 output tokens, DeepSeek V4 Pro costs $0.0172 and GPT-6.1 Sol costs $0.0300, so DeepSeek V4 Pro is about 1.7× cheaper. For output-heavy work (500 in / 4K out), DeepSeek V4 Pro is cheaper.
How much do 1,000 calls cost on each?
At 2,000 input / 500 output tokens per call, 1,000 calls cost $4.62 on DeepSeek V4 Pro and $9.00 on GPT-6.1 Sol. At 20,000 input / 1,000 output, they cost $30.36 and $50.00.
How do cached input prices compare?
DeepSeek V4 Pro: $0.044 per 1M cached input tokens. GPT-6.1 Sol: $0.1 per 1M cached input tokens.
What does a long-context request cost?
A 100,000-input / 1,000-output request costs $0.1360 on DeepSeek V4 Pro and $0.2100 on GPT-6.1 Sol at standard rates. DeepSeek V4 Pro lists a 1M context window; GPT-6.1 Sol lists 1.05M.
Is there a batch discount, and what should I check before switching?
Batch pricing is published for GPT-6.1 Sol 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, DeepSeek V4 Pro or GPT-6.1 Sol?
On cost alone, a 60,000-token agent context with a 4,000-token answer costs $0.0950 on DeepSeek V4 Pro and $0.1600 on GPT-6.1 Sol 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.