Claude Opus 5 API: Fable 5-Class at Half Price, Complete Guide

TL;DR — Anthropic shipped Claude Opus 5 on July 24, 2026 at $5/$25 — the same price as Opus 4.8 and half of Fable 5. The pitch: near-Fable-5 intelligence at half the cost. Model ID is claude-opus-5, 1M context, 128K max output, knowledge cutoff May 2026. The three API changes that matter for migration: adaptive thinking is now on by default, disabling thinking is capped at high effort (disabled + xhigh/max = 400), and the cache minimum drops to 512 tokens. Everything else — pricing, context, sampling-param rejection, no prefill — is unchanged from 4.8. Tested on ofox with our own key (2026-07-25).

What Anthropic Shipped

Claude Opus 5 is Anthropic’s new recommended default for complex agentic coding and enterprise work. The model ID is claude-opus-5 — a fixed, dateless snapshot, no -20260724 suffix to track. It carries the full 1M-token context window by default, 128K max output tokens (up to 300K through the Message Batches API with the output-300k-2026-03-24 beta header), and a May 2026 knowledge cutoff.

The headline is the price-to-capability ratio. Anthropic describes Opus 5 as “a thoughtful and proactive model that comes close to the frontier intelligence of Claude Fable 5 at half the price.” Fable 5 lists at $10/$50 per million; Opus 5 lists at $5 input / $25 output — the same rate Opus 4.8 has carried since May, and half of Fable 5’s. Fast Mode runs the same model at up to 2.5x output speed for double the base cost.

To be clear about the tiering: Fable 5 is still Anthropic’s most capable model. Opus 5 is the model you reach for when you want most of that capability without paying the Fable premium.

Benchmarks: Anthropic’s Own Numbers

Every figure below is Anthropic internal, unverified: vendor launch numbers, not third-party results. Treat them as a best case and run your own evals before you commit. The pattern is at least consistent across a wide set:

Every number in this table is Anthropic internal, unverified — Anthropic’s own launch benchmarks. We did not independently reproduce them.

The through-line is agentic and computer-use work — long-horizon coding, desktop control, real-task automation — where Opus 5 either matches Fable 5 for far less money or clears the rest of the field outright. The one honest caveat Anthropic states up front: Opus 5 stays behind Mythos 5 on cybersecurity-exploitation and biology-research tasks, which are handled by a separate model line for defensive use. If your workload isn’t in those two domains, that gap doesn’t touch you.

For an independent cross-check on how the last generation actually held up on real work, our Opus 4.8 release breakdown walks through the third-party GDPval-AA leaderboard in detail.

What Changed from Opus 4.8 (the part that breaks code)

Opus 5 keeps the same request surface as 4.8 in most respects, but three changes will surprise you if you swap the model string and ship.

1. Adaptive thinking is on by default. On Opus 4.8, a request with no thinking field ran without thinking. On Opus 5, the same request now runs with adaptive thinking. Because max_tokens is a hard cap on total output — thinking plus visible response — a workload tuned for no-thinking 4.8 can now truncate. Revisit max_tokens, or pass thinking: {"type": "disabled"} to keep the old behavior. (Verified on ofox, 2026-07-25: a request to claude-opus-5 with no thinking field returns a thinking block.)

2. Disabling thinking is capped at high effort. This is the new 400 nobody expects. You can still turn thinking off, but only at effort high or below. Combine thinking: {"type": "disabled"} with effort xhigh or max and the request is rejected:

# Rejected on Opus 5 — 400 error
client.messages.create(
    model="claude-opus-5",
    max_tokens=16000,
    thinking={"type": "disabled"},
    output_config={"effort": "xhigh"},   # xhigh/max + disabled = 400
    messages=[{"role": "user", "content": "..."}],
)

# Fix: either re-enable thinking at high effort...
output_config={"effort": "xhigh"}         # thinking on by default
# ...or keep it off and drop to high or below
thinking={"type": "disabled"}, output_config={"effort": "high"}

3. The cache minimum drops to 512 tokens (from 1,024 on 4.8). Short system prompts that were too small to cache before now create cache entries with no code change — a small, free win on prompt-caching bills. (Verified on ofox, 2026-07-25: a ~530-token cached prefix on claude-opus-5 creates a cache entry; ~515 tokens does not.)

Two more worth knowing: mid-conversation tool changes are now possible without invalidating the prompt cache (beta header mid-conversation-tool-changes-2026-07-01), and Opus 5 ships cybersecurity safety classifiers, so you should handle stop_reason: "refusal" and can opt into automatic model fallbacks (fallbacks: "default", beta header server-side-fallback-2026-07-01).

What is not new: pricing, the 1M context window, 128K output, and the hard rejections carried over from 4.8 — non-default temperature/top_p/top_k, manual budget_tokens, and assistant prefill all still return 400. Priority Tier is still unsupported.

Effort and Prompting Notes

Opus 5 supports the full effort ladder — low, medium, high, xhigh, max — and defaults to high on the Claude API and Claude Code. Anthropic’s guidance is to run a fresh effort sweep rather than carry over 4.8 settings, and to test max for capability-critical work (set a large max_tokens, starting around 64K, when you run xhigh or max).

Two behavior shifts are worth a prompt review. Opus 5 verifies its own work without being told, so explicit “double-check your output” instructions now cause over-verification — remove them. And its default responses run longer than 4.8’s; lowering effort trims thinking but not visible length, so prompt for conciseness or a target length directly if you need shorter output.

How to Call Opus 5 Through ofox

The model ID on ofox.ai is anthropic/claude-opus-5, on the same OpenAI-compatible endpoint as every other model — no separate Anthropic account, no separate billing. Point your existing OpenAI SDK at the ofox base URL and change two strings:

from openai import OpenAI

client = OpenAI(base_url="https://api.ofox.io/v1", api_key="your-ofox-key")

response = client.chat.completions.create(
    model="anthropic/claude-opus-5",
    messages=[{"role": "user", "content": "Refactor this module for testability..."}],
)

To use adaptive thinking and the effort parameter, call the Anthropic-native endpoint instead — the model ID drops the provider prefix:

import anthropic

client = anthropic.Anthropic(
    base_url="https://api.ofox.io/anthropic",
    api_key="your-ofox-key",
)

response = client.messages.create(
    model="claude-opus-5",
    max_tokens=8000,
    thinking={"type": "adaptive"},
    output_config={"effort": "high"},   # low | medium | high | xhigh | max
    messages=[{"role": "user", "content": "Audit this service for race conditions..."}],
)

(Both call paths above verified on ofox, 2026-07-25: the OpenAI-compatible route returns a normal completion, and the Anthropic-native route with thinking: {"type": "adaptive"} plus effort returns a thinking block.)

Running through one gateway also makes the migration question empirical: put the same prompts through Opus 5, Opus 4.8, and Fable 5 on one key and compare quality and token counts on your workload before switching production.

Opus 5 vs Fable 5 vs Sonnet 5: Which to Pick

• Claude Opus 5 ($5/$25) — the new default for agentic coding, computer use, and long-horizon enterprise tasks. Best capability-per-dollar in the lineup right now.
• Claude Fable 5 ($10/$50) — the ceiling. Reach for it only when a task genuinely needs the most capable model and cost is secondary; on many agentic benchmarks Opus 5 is within a rounding error at half the price.
• Claude Sonnet 5 ($3/$15) — the speed-and-cost tier. Near-Opus quality on coding for high-volume or latency-sensitive work where you don’t need Opus-level reasoning depth.

If you’re on Opus 4.8 today, this is a clean migration: same price, higher scores, one model-string change plus the three gotchas above.

Verdict

Opus 5 is another Opus upgrade with no asterisk on price — same $5/$25, materially higher scores across coding and computer-use, and Anthropic’s claim of near-Fable-5 capability at half the cost holds up across a broad benchmark set. The caveats are narrow and honest: the benchmark numbers are Anthropic internal and unverified (run your own evals), disabling thinking now conflicts with high effort levels, and default output runs longer so watch your max_tokens and latency budgets.

For new projects, start on Opus 5. For anything in production on 4.8, change the model string, revisit max_tokens, audit any request that disables thinking, and ship.

---

Related: Claude Opus 4.8 release breakdown — the predecessor and its independent leaderboard result. Fable 5 vs Opus 4.8 vs GPT-5.5 on SWE-bench — where the flagship tier lands on coding. Best AI Model for Coding 2026 and Best AI Model for Agents 2026 — where Opus 5 fits the wider landscape.