Model comparison
Grok 4.7 vs Claude Opus 5.5: cheaper tokens, smaller window
Grok 4.7 costs $2.30 on a cached coding session against $4.40 on Claude Opus 5.5, but its window is 500K, not 1M, and prompts from 200K cost double.
· Prices as of September 28, 2026
Grok 4.7
xAI · Released September 21, 2026
xAI's top model for coding and knowledge work, which xAI says works longer on hard tasks and checks its own work more carefully.
Grok 4.7 facts and comparisonsClaude Opus 5.5
Anthropic · Released September 22, 2026
Anthropic's recommended starting model for most work, built for long-running agentic coding. Anthropic says it matches Claude Fable 5.1 on most work at a much lower price.
Claude Opus 5.5 facts and comparisons
The short answer
On the example agentic coding session Grok 4.7 costs $2.30 against $4.40 on Claude Opus 5.5, and output-heavy work costs 3.2x as much on Opus 5.5 because its output rate is $20 per million against $6. Opus 5.5 keeps a 1M context window at standard rates, while Grok 4.7 stops at 500K and doubles every rate once a prompt reaches 200K tokens. Grok 4.7 suits cost-sensitive work in Cursor or GitHub Copilot, and Opus 5.5 suits long, sprawling jobs in Claude Code, which Anthropic names as a particular strength.
Choose Grok 4.7 if
- You want the lower list prices: $2 input and $6 output per million tokens, against $4 and $20 on Opus 5.5.
- Your work is output-heavy, where the gap is widest: $0.54 against $1.72 for the example generation.
- You code in Cursor, which lists Grok 4.7 as trained jointly by Cursor and xAI.
- Your prompts stay well under 200K tokens, below the line where every Grok 4.7 rate doubles.
Choose Claude Opus 5.5 if
- You work in Claude Code, where Opus 5.5 is the default model on Pro, Max, Team, and Enterprise plans and with an Anthropic API key.
- Your backlog includes codebase-wide migrations or audits, the long, sprawling kind of job Anthropic highlights for Opus 5.5.
- You need more than 500K tokens of context, or want the full 1M window billed at standard rates.
- Your sessions reread their cached context far more than the example session does, since an Opus 5.5 cache hit costs $0.20 per million against $0.50 on Grok 4.7.
Side by side
Specs and prices
| Fact | Grok 4.7 | Claude Opus 5.5 |
|---|---|---|
| Maker | xAI | Anthropic |
| API model id | grok-4.7 | claude-opus-5-5 |
| Released | September 21, 2026 | September 22, 2026 |
| Status | Current | Current |
| Context window | 500K tokens | 1M tokens |
| Max output | Not published | 128K tokens |
| Open weights | No | No |
| Input, per 1M tokens | $2 | $4 |
| Cache hit, per 1M | $0.50 | $0.20 |
| Cache write, per 1M | $2 (same as input) | $5 (5-minute), $8 (1-hour) |
| Output, per 1M tokens | $6 | $20 |
| Runs in | Cursor, OpenCode, OpenRouter, and GitHub Copilot | Claude Code, Cursor, OpenCode, OpenRouter, and GitHub Copilot |
Standard API rates in US dollars, as published by each maker (Grok 4.7: September 28, 2026; Claude Opus 5.5: September 26, 2026). Batch and priority tiers, taxes, and subscription plans are not included. Grok 4.7: Once a prompt reaches 200K tokens, every token in the request costs $4 input, $1 cached, and $12 output per million. The US regional endpoint costs 10% more. Claude Opus 5.5: The full 1M context window is billed at standard rates. Fast mode, a research preview on the Claude API, costs $8 input and $40 output per million tokens.
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 | Grok 4.7 | Claude Opus 5.5 |
|---|---|---|
| Agentic coding session, 100K input, 400K written to cache, 2M read from cache, 50K output | $2.30 | $4.40 |
| Large one-off review, 150K input with no cache hits, 10K output | $0.36 | $0.80 |
| Output-heavy generation, 30K input, 80K output | $0.54 | $1.72 |
| A month of sessions, 110 sessions: 5 a day, 22 working days | $253.00 | $484.00 |
| Where the session’s cost goes | ||
| Cache writes | $0.80 | $2.60 |
| Cache reads | $1.00 | $0.40 |
| Uncached input | $0.20 | $0.40 |
| Output | $0.30 | $1.00 |
- caching saves on the session with Grok 4.7 (57%)
- $3.00
- caching saves on the session with Claude Opus 5.5 (60%)
- $6.60
Where the Grok 4.7 and Claude Opus 5.5 price gap comes from
Grok 4.7 lists $2 per million input tokens and $6 per million output tokens on xAI's API. Claude Opus 5.5 lists $4 and $20 on Anthropic's. Input is 2x dearer on Opus 5.5 and output 3.3x, so the output-heavy generation shows the widest spread: $0.54 on Grok 4.7 against $1.72, a 3.2x difference. The large one-off review, which is mostly input, lands at $0.36 against $0.80.
Cache hits pull the other way. Anthropic prices an Opus 5.5 hit at 0.05x input, or $0.20 per million, while xAI charges $0.50 for a Grok 4.7 hit, 25% of its input price. The example session reads 2M tokens from the cache, which costs $1.00 on Grok 4.7 and $0.40 on Opus 5.5. Reads are the largest line in Grok 4.7's session, at 43% of its total.
Cache writes decide the rest. xAI lists no fee for writing the cache, so the 400K written tokens cost ordinary input, $0.80. Anthropic bills 5-minute writes at 1.25x input and 1-hour writes at 2x, which adds up to $2.60 on Opus 5.5, or 59% of its session. The session ends at $2.30 against $4.40, a 1.9x gap, and at $253.00 against $484.00 over 110 sessions a month.
How the 200K line and the 500K window change the math
Grok 4.7 has a 500K context window, half the 1M on Opus 5.5. xAI also sets a long-context rate: once a prompt reaches 200K tokens, every token in that request costs $4 input, $1 cached, and $12 output per million. The higher rate covers the whole request, not only the tokens past the line.
In that tier Grok 4.7's input matches the $4 that Opus 5.5 charges, and its cached rate is several times the Opus 5.5 hit price. Output still costs less, $12 against $20. Anthropic bills the full 1M window on Opus 5.5 at standard rates, so a request that loads a large repository or a long log costs the same per token as a short one.
The example session keeps each request under 200K tokens, so none of this shows in the cost table. An agent whose context stays above 200K on every turn would pay Grok 4.7's higher rates on each of those turns. Two smaller points: xAI's US regional endpoint costs 10% more, and Opus 5.5 writes up to 128K tokens of output, while xAI does not publish a maximum output for Grok 4.7.
What xAI and Anthropic say each model is for
xAI calls Grok 4.7 "our most capable model for coding and knowledge work." It says the model works longer on difficult tasks and checks its own work more carefully, and that it sits on a new, larger base model trained with a longer reinforcement-learning run on many-hour tasks.
Anthropic positions Opus 5.5 as its recommended starting model for most work and says "it performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5." It names agentic coding, computer use, and knowledge work as strengths, along with long, sprawling jobs like codebase-wide migrations and audits.
Each has a faster option at double the price. Opus 5.5 fast mode, a research preview on the Claude API, costs $8 input and $40 output per million tokens. Cursor lists a faster Grok 4.7 Fast at double the rates. Neither changes the standard rates in the tables above.
Which coding tools run Grok 4.7 and Claude Opus 5.5
Both models are offered in Cursor, OpenCode, OpenRouter, and GitHub Copilot, which carry models from several makers. Opus 5.5 also runs in Claude Code, Anthropic's own coding agent. There the main conversation uses the 1-hour cache on a Claude subscription and the 5-minute cache with an API key, so an API-key user pays less for writes than this session assumes.
The cost table uses xAI's and Anthropic's own API prices, so read every figure as an API-equivalent estimate at published rates. Subscription plans are priced differently. Tokenizers differ too: every Claude model from 4.7 on counts about 30% more tokens than earlier Claude models for the same text, and the same prompt will not produce the same token count on both.
EveryToken prices Opus 5.5 at Anthropic's rates from Claude Code, Cursor, and OpenCode history, and prices Grok 4.7 only when you use it through OpenRouter, from OpenRouter's catalog.
Prompt caching
How each maker bills cached tokens
xAI
The xAI API caches repeated prompt prefixes automatically. Sending the same conversation id with each request raises the hit rate.
xAI lists no fee for writing the cache. A cache hit costs $0.50 per million tokens on Grok 4.7 and $0.20 on Grok Build 0.1.
Source: xAI docs: Prompt caching
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
Your own numbers
See what Grok 4.7 and Claude Opus 5.5 really cost you.
everyaitoken reads your OpenRouter, Claude Code, Cursor, and OpenCode history on your Mac and prices every request at API rates, with what caching saved or cost. $9 once.
FAQ
Questions
Is Grok 4.7 cheaper than Claude Opus 5.5?
Per token, mostly yes: $2 input and $6 output per million against $4 and $20. Cache hits are the exception, at $0.50 on Grok 4.7 and $0.20 on Opus 5.5. On the example session Opus 5.5 still costs 1.9x as much, $4.40 against $2.30.
What happens when a Grok 4.7 prompt reaches 200K tokens?
xAI bills every token in that request at $4 input, $1 cached, and $12 output per million, double the standard rates. Grok 4.7's window ends at 500K. Opus 5.5 bills its whole 1M window at standard rates.
Which tools offer both Grok 4.7 and Claude Opus 5.5?
Cursor, OpenCode, OpenRouter, and GitHub Copilot list both. Claude Code, Anthropic's own coding agent, runs Opus 5.5 and uses it as the default model on Pro, Max, Team, and Enterprise plans.
Do both models have a fast mode?
Opus 5.5 has fast mode, a research preview on the Claude API at $8 input and $40 output per million tokens. For Grok 4.7, Cursor lists a faster Grok 4.7 Fast at double the rates.
Sources
- xAI docs: Grok 4.7
- xAI: Grok 4.7
- OpenRouter: Grok 4.7
- OpenCode docs: Zen
- Cursor docs: Models
- GitHub Docs: Supported AI models in Copilot
- Anthropic: Pricing
- Anthropic docs: Claude Opus 5.5
- Anthropic: Claude Opus 5.5
- Anthropic docs: Fast mode
- Claude Code docs: Model configuration
- Cursor docs: Claude Opus 5.5
- OpenRouter: Claude Opus 5.5
- OpenCode docs: Zen
- xAI docs: Prompt caching
- Anthropic: Prompt caching