Claude Opus 4.6 extended the Opus line to long-context work: compared with Opus 4.5 it grows the context window from 200K to 1M tokens, doubles max output to 128K tokens, and introduces adaptive thinking alongside extended thinking, all at unchanged $5/$25-per-million-token pricing.
- Input
- text image $5/M
- Output
- text $25/M
- Cache read
- $0.5/M
- Context
- 1M
- Knowledge cutoff
- 2025-05
Benchmarks
Vendor-published: Alibaba (Qwen) Anthropic ByteDance DeepSeek Google MiniMax Moonshot OpenAI Tencent Z.ai
Price in context
Where the price sits among 60 comparable models
The bar shows how this model’s price compares with every other model of the same kind on Synthorai. The cheapest and the most expensive are named at each end. These are base rates; batch, region and cache-write discounts are on the pricing page.
Specs & limits
Tokens
| Context window (vendor spec) | 1,000,000 |
|---|---|
| Max output (vendor spec) | 128,000 |
| Knowledge cutoff | 2025-05 |
| Training data through | 2025-08 knowledge cutoff 2025-05 |
Prompt caching
| How it caches | explicit (opt-in) |
|---|---|
| Min prefix | 4,096 provider default is 1,024 |
| Lifetime | 5m default, 1h option |
| Write cost | 1.25x (5m) / 2x (1h) |
Thinking
| Vendor control | thinking.type + output_config.effort (budget_tokens still accepted, deprecated) |
|---|---|
| Accepted values | thinking.type adaptive · enabled (deprecated) · disabled; effort low · medium · high · max (no xhigh) |
| Default | thinking off; effort high applied when the request sets nothing |
| Can be turned off | Yes |
| Thinking behaviour | Adaptive thinking interleaves between tool calls automatically with no beta header, while manual "enabled" mode has no interleaved thinking at all on this model; earlier turns' thinking blocks are kept in context and billed as input. |
| Parameter | reasoning_effort |
| Values | minimal · low · medium · high the gateway's parameter surface - the vendor mapping above applies |
Model
| Modalities | text + image → text |
|---|
- First Opus with adaptive thinking recommended
- assistant prefill not supported
- 1M context GA since 2026-03-13
- fast mode removed 2026-06-29
One prompt, measured through the gateway
Claude Opus 4.6 passed · 3 sentences
When reasoning is enabled, the model generates extensive internal chain-of-thought tokens that are produced—and billed—before the user-visible answer even begins, often multiplying total token consumption by 5–10× per request. Because our pricing is strictly per-token for both input and output, those hidden "thinking" tokens hit the invoice just as hard as the final response, even though end users never see them. To bring costs back in line, we can tune the reasoning effort parameter, cache common prompts, and route only genuinely complex queries to the reasoning model while handling routine tasks with a lighter, cheaper one.
out 132 tok latency 6.3 s
Instruction following (exactly three sentences - countable), audience shaping (CFO register), and the hidden-thinking billing gap the token meters below expose.
Claude Opus 4.6 passed · 8/8 cases
```python def longest_run(xs): if not xs: return 0 best = 1 cur = 1 for i in range(1, len(xs)): if xs[i] == xs[i - 1]: cur += 1 best = max(best, cur) else: cur = 1 return best ``` The original code had two related bugs stemming from the fact that it counted additional consecutive elements (starting from 0) rather than the total length of a run (starting from 1): first, `best` and `cur` were initialized to 0 instead of 1, which meant a single-element list (where the loop never executes) returned 0 instead of the correct answer 1; second, when a mismatch was found, `cur` was reset to 0 instead of 1 (failing to count the current element as the start of a new run), which meant that if the longest run occurred at the very end of the list, the comparison `best = max(best, cur)` on the next matching step would be w […]
out 277 tok latency 7.0 s
Whether the fix is actually correct (runnable), explanation density, and token efficiency on a bounded task.
Claude Opus 4.6 passed · 5/5 fields, guidance "withheld pending the Q4 close"
```json { "company": "Meridian Robotics", "quarter": "September quarter", "revenue_usd_m": 412, "growth_pct": 11, "guidance": "withheld pending the Q4 close" } ```
out 73 tok latency 2.6 s
Schema obedience (no invented fields), hallucination pressure (guidance is explicitly withheld), and structured-output path differences.
Claude Opus 4.6 passed · 118 words, 0 banned, 1 question
**Introducing Cross-Provider Prompt Caching** Why pay for the same computation twice? Our new API gateway feature intelligently caches prompt responses across multiple LLM providers, reducing latency by up to 60% and cutting costs significantly for repeated or similar queries. When your application sends a prompt that matches a cached response—regardless of whether it originally came from OpenAI, Anthropic, or Cohere—the gateway serves the stored result instantly. You maintain full control over cache duration, invalidation rules, and provider-specific overrides through a simple configuration dashboard. The feature supports both exact and semantic matching, letting teams optimize for speed or flexibility based on their use case. Available today on all Pro and Enterprise plans with no additional setup required.
out 163 tok latency 6.1 s
Constraint obedience (word budget, banned-word list, the single question), style fingerprint, and length control.
Use Claude Opus 4.6 in 30 seconds
OpenAI-compatible: swap the base_url, keep your SDK. POST /v1/chat/completions
from openai import OpenAI
client = OpenAI(
base_url="https://synthorai.io/v1",
api_key="sk-syn-...",
)
resp = client.chat.completions.create(
model="claude-opus-4-6",
messages=[{"role": "user", "content": "Summarize this diff"}],
reasoning_effort="medium",
)
print(resp.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://synthorai.io/v1",
apiKey: "sk-syn-...",
});
const resp = await client.chat.completions.create({
model: "claude-opus-4-6",
messages: [{ role: "user", content: "Summarize this diff" }],
reasoning_effort: "medium",
});
console.log(resp.choices[0].message.content);curl https://synthorai.io/v1/chat/completions \
-H "Authorization: Bearer sk-syn-..." \
-H "Content-Type: application/json" \
-d '{
"model": "claude-opus-4-6",
"messages": [{"role": "user", "content": "Hello"}],
"reasoning_effort": "medium"
}'package main
import (
"context"
"fmt"
"github.com/openai/openai-go/v3"
"github.com/openai/openai-go/v3/option"
)
func main() {
client := openai.NewClient(
option.WithBaseURL("https://synthorai.io/v1"),
option.WithAPIKey("sk-syn-..."),
)
resp, _ := client.Chat.Completions.New(context.TODO(), openai.ChatCompletionNewParams{
Model: "claude-opus-4-6",
Messages: []openai.ChatCompletionMessageParamUnion{
openai.UserMessage("Summarize this diff"),
},
ReasoningEffort: openai.ReasoningEffortMedium,
})
fmt.Println(resp.Choices[0].Message.Content)
}import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.chat.completions.*;
import com.openai.models.ReasoningEffort;
OpenAIClient client = OpenAIOkHttpClient.builder()
.baseUrl("https://synthorai.io/v1")
.apiKey("sk-syn-...")
.build();
ChatCompletion resp = client.chat().completions().create(
ChatCompletionCreateParams.builder()
.model("claude-opus-4-6")
.addUserMessage("Summarize this diff")
.reasoningEffort(ReasoningEffort.MEDIUM)
.build());
System.out.println(resp.choices().get(0).message().content().orElse(""));About Claude Opus 4.6
- Anthropic launched it as its most intelligent model for complex agentic tasks and long-horizon work, and the 1M window moved from beta to generally available at standard pricing a month later.
- Vision input, tool use, and prompt caching are supported, and the Batch API offers an extended 300K-output beta for this model.
- This is also the release where compaction first shipped in beta, giving long agent sessions a way to keep working past the window.
- Both thinking modes are accepted, with extended thinking now deprecated but still functional and neither type rejected; effort covers low through max without xhigh and defaults to high.
- Two behaviours catch migrations: prefilling the assistant message is not supported from this generation onward, and the fast-mode research preview was removed in June 2026, after which requests simply run at standard speed and standard rates instead of erroring.
- Prompt caching requires a 4,096-token minimum prefix.
- It is now listed as a legacy model behind Opus 4.7 and 4.8.
- Synthorai makes Claude Opus 4.6 callable from any OpenAI-compatible client.
FAQ
Is the Claude Opus 4.6 API free to try?
Yes: new accounts get 10 trial calls and up to $1 in free credit, no card required. At $5/M input tokens, that credit alone covers roughly 24 requests of ~8K tokens against Claude Opus 4.6.
What is Claude Opus 4.6 best at?
Context window grown from 200K to 1M; doubled max output to 128K tokens; extended 300K-output batch beta. See the About section for the full picture from the vendor's own release notes.
How much does Claude Opus 4.6 cost?
Claude Opus 4.6 costs $5 per million input tokens and $25 per million output tokens on Synthorai. That is the provider's list price, with no platform markup. Cached input tokens bill at $0.5/M.
Does Claude Opus 4.6 support prompt caching?
Yes, via opt-in: mark stable prefixes with cache_control breakpoints. Cached input tokens bill at $0.5/M vs $5/M uncached; prompts need a 4,096-token stable prefix to cache (TTL 5m default, 1h option). Prompt caching guide →
How do I get access to Claude Opus 4.6?
Point your existing OpenAI SDK at base_url="https://synthorai.io/v1", set model="claude-opus-4-6", and you're done. One API key covers every model on the gateway.
What is Claude Opus 4.6's knowledge cutoff?
Claude Opus 4.6's knowledge cutoff is 2025-05, per the vendor's official documentation (as of 2026-07-09).
Related models
Compare
Every value on this page is transcribed from the vendor's own documentation, linked above, and carries the date it was checked. Prices are compared across the catalogue; specification values that vendors define differently are shown with the difference stated rather than charted. Nothing here is measured by us, and nothing is scored.