For coding agents, GPT-5.5 costs 2.0x as much as GPT-5.4 per request ($0.21 vs $0.10 on our measured traffic), because GPT-5.5 lists at $5 input, $0.50 cached input and $30 output per million tokens against $2.50, $0.25 and $15 for GPT-5.4, and both charge more for long prompts.
Run GPT-6.1 Sol in your agent
Or let the router send each step to the cheapest model that can do it, on providers we vet for price, uptime and speed. Plans from $19 a month.
| Cost at list price | GPT-5.4 | GPT-5.5 |
|---|---|---|
| Prompt under 100K tokens50% of requests, 87% of input cached | $0.040 | $0.081 |
| Prompt of 100K to 272K39% of requests, 88% of input cached | $0.099 | $0.20 |
| Prompt over 272K tokens11% of requests, 86% of input cached | $0.43 | $0.86 |
| Average request | $0.10 | $0.21 |
| 100 requests | $10.49 | $20.99 |
| Median session16 requests; a session ends after 10 idle minutes | $1.15 | $2.30 |
| A month of regular usemedian of 20 users active on 5+ days | $148 | $296 |
Each logged request priced at the vendor’s list rates for its own token counts, including cache reads and any higher rate for long prompts. Average output: 698 tokens per request.
| Artificial Analysis | GPT-5.4 | GPT-5.5 |
|---|---|---|
| Intelligence Index10 evaluations, 0 to 100 | 39 (estimate) | 38 |
| Terminal-Bench 4.0agentic coding in a terminal | n/a | 14.6% |
| Output speedmedian | 95 tokens/s | 105 tokens/s |
| Time to first answer tokenincludes thinking time | 3.3 min | 71 s |
Source: Artificial Analysis (artificialanalysis.ai). Measured variants: GPT-5.4 (Xhigh), speed on OpenAI; GPT-5.5 (Xhigh), speed on OpenAI. Fetched October 13, 2026. Artificial Analysis runs Terminal-Bench 4.0 in its own setup, so its scores differ from the tbench.ai leaderboard.
GPT-5.5 is superseded: GPT-6.1 Sol costs $0.067 per request. GPT-5.4 scores as well or better on the benchmarks above, so the cheaper request is not bought with much lower ability.
Pick a winner in each dimension. You can change your vote any time.
Correctness and depth of what it produces
Tool calls, instruction-following, and multi-step tasks
Tokens per second and time-to-first-token
How much capability you get per dollar
Consistent results: fewer refusals, loops and format breaks
The agent matters as much as the model. Compare AI agents →
Current Claude, GPT, Gemini and open models run on Standard Compute