Model comparison
Claude Opus 4.8 vs GPT-5.5: which costs less to code with?
Claude Opus 4.8 and GPT-5.5 share a $5 input price, yet GPT-5.5 costs less on cached sessions and Opus 4.8 less on output-heavy work. Where the crossover falls.
· Prices as of September 28, 2026
Claude Opus 4.8
Anthropic · Released May 28, 2026 · Previous generation
An Opus upgrade over 4.7 focused on judgment and collaboration. Anthropic still recommends it for cybersecurity work that needs reduced guardrails.
Claude Opus 4.8 facts and comparisonsGPT-5.5
OpenAI · Released April 23, 2026 · Previous generation
OpenAI's April 2026 flagship for coding and professional work, now a previous generation that stays in the API.
GPT-5.5 facts and comparisons
The short answer
It depends on the shape of the work. GPT-5.5 costs less on the example agentic coding session, $5.00 against $6.00, because OpenAI bills cache writes as ordinary input while Anthropic charges up to 2x, but Claude Opus 4.8 costs less on output-heavy work because its output is $25 per million against $30. Both are previous-generation models, and GPT-5.5 leaves ChatGPT and Codex sign-in on October 14, 2026.
Choose Claude Opus 4.8 if
- Your work is output-heavy, such as generating large files, where the example generation costs $2.15 on Opus 4.8 against $2.55.
- Your team does cybersecurity work that needs reduced guardrails, which Anthropic still recommends Opus 4.8 for.
- You work in Claude Code, where Opus 4.8 remains available as a legacy model.
- Your prompts run past 272K input tokens, where GPT-5.5 bills 2x input and 1.5x output for the full session.
Choose GPT-5.5 if
- Your sessions write a lot to the cache, which GPT-5.5 bills at its ordinary $5 input rate with no premium.
- You call OpenAI with an API key, where GPT-5.5 stays available after it leaves ChatGPT and Codex sign-in.
- OpenAI's claim of strong results with fewer reasoning tokens than earlier models at the same effort matters to your costs.
Side by side
Specs and prices
| Fact | Claude Opus 4.8 | GPT-5.5 |
|---|---|---|
| Maker | Anthropic | OpenAI |
| API model id | claude-opus-4-8 | gpt-5.5 |
| Released | May 28, 2026 | April 23, 2026 |
| Status | Previous generation | Previous generation |
| Context window | 1M tokens | 1.05M tokens |
| Max output | 128K tokens | 128K tokens |
| Open weights | No | No |
| Input, per 1M tokens | $5 | $5 |
| Cache hit, per 1M | $0.50 | $0.50 |
| Cache write, per 1M | $6.25 (5-minute), $10 (1-hour) | $5 (same as input) |
| Output, per 1M tokens | $25 | $30 |
| Runs in | Claude Code, Cursor, OpenCode, OpenRouter, and GitHub Copilot | Codex, Cursor, OpenCode, OpenRouter, and GitHub Copilot |
Standard API rates in US dollars, as published by each maker (Claude Opus 4.8: September 26, 2026; GPT-5.5: September 28, 2026). Batch and priority tiers, taxes, and subscription plans are not included. Claude Opus 4.8: The full 1M context window is billed at standard rates. Fast mode, a research preview on the Claude API, costs $10 input and $50 output per million tokens. GPT-5.5: Prompts over 272K input tokens cost 2x for input and 1.5x for output, for the full session.
Cost
What the same work costs
The same token counts, priced at each model’s published rates. Cached tokens are billed at each maker’s cache prices, so the session shows what caching is worth on each model.
Real sessions differ: the two models count the same code as different numbers of tokens, and reasoning settings change how much each one writes. Your own history is the real test.
| Workload | Claude Opus 4.8 | GPT-5.5 |
|---|---|---|
| Agentic coding session, 100K input, 400K written to cache, 2M read from cache, 50K output | $6.00 | $5.00 |
| Large one-off review, 150K input with no cache hits, 10K output | $1.00 | $1.05 |
| Output-heavy generation, 30K input, 80K output | $2.15 | $2.55 |
| A month of sessions, 110 sessions: 5 a day, 22 working days | $660.00 | $550.00 |
| Where the session’s cost goes | ||
| Cache writes | $3.25 | $2.00 |
| Cache reads | $1.00 | $1.00 |
| Uncached input | $0.50 | $0.50 |
| Output | $1.25 | $1.50 |
- caching saves on the session with Claude Opus 4.8 (56%)
- $7.75
- caching saves on the session with GPT-5.5 (64%)
- $9.00
Same input price, different cache and output rates
Claude Opus 4.8 and GPT-5.5 both charge $5 per million input tokens and $0.50 per million for a cache hit. They split on two lines. Output costs $25 per million on Opus 4.8 and $30 on GPT-5.5, 20% more. Cache writes cost $6.25 or $10 on Opus 4.8, for 5-minute and 1-hour writes, while GPT-5.5 adds no charge for writing the cache and bills those tokens at its $5 input rate.
Which line dominates decides which model costs less. The example agentic session writes 400K tokens to the cache and produces 50K of output, so the write premium matters most: writes cost $3.25 on Opus 4.8 against $2.00 on GPT-5.5, and the session costs $6.00 against $5.00. The output-heavy generation turns that around, $2.15 on Opus 4.8 against $2.55, and the large one-off review lands close, $1.00 against $1.05.
Over 110 sessions a month, the session math adds up to $660.00 on Opus 4.8 and $550.00 on GPT-5.5 at API rates, a $110.00 gap. Caching saves 64% on GPT-5.5 against sending the same tokens uncached, and 56% on Opus 4.8, where the write premium takes back part of the discount.
Long sessions, limits, and GPT-5.5 leaving Codex sign-in
GPT-5.5 has a long-context price tier that Opus 4.8 lacks. Once a prompt goes over 272K input tokens, GPT-5.5 charges 2x for input and 1.5x for output for the full session. Opus 4.8 bills its whole 1M context window at standard rates. Both write up to 128K tokens of output, and GPT-5.5's window is 1.05M.
GPT-5.5 leaves ChatGPT and Codex sign-in on October 14, 2026, and stays in the API, so anyone using it through a ChatGPT plan in Codex will need another model after that date. Opus 4.8 is a legacy Claude model that stays available, and it is listed in Claude Code, Cursor, OpenRouter, OpenCode, and GitHub Copilot.
Caching is also controlled differently. GPT-5.5 caches automatically only, without the explicit breakpoints newer OpenAI models allow. On Claude, one top-level cache_control field places a breakpoint and moves it as the conversation grows, or you mark up to 4 blocks yourself, and Claude Code manages caching for you.
How Anthropic and OpenAI describe these two models
OpenAI launched GPT-5.5 in April 2026 as "A new class of intelligence for coding and professional work." It credits the model with strong results using fewer reasoning tokens than earlier models at the same effort, with complex coding that needs planning, tool use, codebase navigation, verification, and multi-step execution, and with precise tool use on large tool surfaces and long-running agent tasks.
Anthropic released Opus 4.8 a month later as an upgrade over Opus 4.7 focused on judgment and collaboration, with the consistency and autonomy to keep working on long-running tasks, and recommends starting it at xhigh effort for coding. Reasoning settings shape cost on both, since they change how much each model writes, and the makers' tokenizers differ, so the same text won't count identically on each.
To see which way the crossover falls for your own mix of cached and output-heavy work, EveryToken prices your local Claude Code and Codex history at each maker's API rates and shows what caching saved or cost per model.
Prompt caching
How each maker bills cached tokens
Anthropic
Claude caches a prompt prefix up to a breakpoint. One top-level cache_control field places the breakpoint automatically and moves it as the conversation grows, or you can mark up to 4 blocks yourself. Claude Code manages caching for you.
A 5-minute cache write costs 1.25x the input price and a 1-hour write costs 2x. A cache hit costs 0.1x input on most models, 0.05x on Claude Opus 5.5, and 0.025x on Claude Fable 5.1 and Claude Mythos 5.1. Every hit restarts the cache lifetime at no charge.
Anthropic's rule of thumb: a 5-minute write pays for itself after one cache read, and a 1-hour write after two.
In Claude Code, the main conversation uses the 1-hour cache on a Claude subscription and the 5-minute cache with an API key. Each model has its own cache, so switching models starts over.
Source: Anthropic: Prompt caching
OpenAI
Prompt caching is on by default. From GPT-5.6 on you can also mark up to four explicit cache breakpoints, while GPT-5.5 and earlier cache automatically only.
From GPT-5.6 on, a cache write costs 1.25x the uncached input price and a cache hit costs 0.1x. GPT-5.5 and earlier add no charge for writing the cache: written tokens are billed as ordinary input, and a hit costs 0.1x on the models compared here.
On GPT-5.6 and later, a cached prefix stays reusable for at least 30 minutes after its last use, and caching starts at 1,024 input tokens.
Source: OpenAI: Prompt caching
Your own numbers
See what Claude Opus 4.8 and GPT-5.5 really cost you.
everyaitoken reads your Claude Code, Cursor, OpenCode, OpenRouter, and Codex history on your Mac and prices every request at API rates, with what caching saved or cost. $9 once.
FAQ
Questions
Is Claude Opus 4.8 or GPT-5.5 cheaper?
That depends on what the work looks like. GPT-5.5 is cheaper on cache-heavy agentic sessions, $5.00 against $6.00 on the example, and Opus 4.8 is cheaper on output-heavy generation, $2.15 against $2.55.
Is there a cache-write charge on GPT-5.5?
Not separately. OpenAI bills written tokens on GPT-5.5 as ordinary input, $5 per million, and a cache hit costs 0.1x input. On Opus 4.8, a 5-minute write costs 1.25x input and a 1-hour write costs 2x.
What happens to GPT-5.5 in Codex in October 2026?
On October 14, 2026, it drops out of ChatGPT and Codex sign-in. Access through the OpenAI API continues after that date.
Are Claude Opus 4.8 and GPT-5.5 still current?
No, both are previous-generation models that stay available. Anthropic's recommended starting model is now Claude Opus 5.5, though it still recommends Opus 4.8 for cybersecurity work that needs reduced guardrails.
Sources
- Anthropic: Pricing
- Anthropic docs: Claude Opus 4.8
- Anthropic: Introducing Claude Opus 4.8
- Anthropic: Claude Opus
- Claude Code docs: Model configuration
- Cursor docs: Claude Opus 4.8
- OpenRouter: Claude Opus 4.8
- OpenCode docs: Zen
- GitHub Docs: Supported AI models in Copilot
- OpenAI: API pricing
- OpenAI docs: GPT-5.5
- OpenAI: Using GPT-5.5
- OpenAI: API changelog
- Codex docs: Models
- Cursor docs: Models and pricing
- OpenRouter: GPT-5.5
- Anthropic: Prompt caching
- OpenAI: Prompt caching