TL;DR: Claude and OpenAI no longer fit a simple “which flagship is cheaper” answer. Claude Sonnet 5 is the first Claude production test at $2/$10 per 1M tokens through August 31, then $3/$15. Opus 4.8 remains the practical premium Claude baseline at $5/$25. OpenAI GPT-5.5 has cached-input pricing and separate short- and long-context lanes.
Anthropic says the June 12 export controls were lifted June 30 and Fable 5 returned globally on native Claude surfaces July 1. Included usage is temporary and plan-specific, cloud re-enablement is rolling out, and Mythos 5 remains restricted to approved organizations. Fable’s restored availability does not make it the default premium lane.
GPT-5.6 Sol, Terra, and Luna are a restricted preview watchlist, not current generally available value routes. OpenAI lists $5/$30, $2.50/$15, and $1/$6 input/output pricing, but access initially requires selection into its trusted-partner cohort. Keep GPT-5.5 and Opus 4.8 as active baselines; see the access-gates analysis.
Use this page for routing decisions, not rankings. Test before moving production workloads.
Current API Price Anchors
Prices are per 1 million tokens.
| Lane | Input | Cached input | Output | Batch input/output | Practical read |
|---|---|---|---|---|---|
| Claude Sonnet 5 | $2 intro; $3 standard | verify current cache rate | $10 intro; $15 standard | verify current pricing docs | Introductory pricing ends Aug 31 |
| Claude Opus 4.8 | $5 | $0.50 cache hits | $25 | $2.50 / $12.50 | Practical premium Claude baseline |
| Claude Fable 5 | $10 | $1 cache hits | $50 | $5 / $25 | Restored guarded escalation; not a default lane |
| GPT-5.6 Sol | $5 | $0.50 cache reads | $30 | not published | Restricted-preview watchlist |
| GPT-5.6 Terra | $2.50 | $0.25 cache reads | $15 | not published | Restricted-preview watchlist |
| GPT-5.6 Luna | $1 | $0.10 cache reads | $6 | not published | Restricted-preview watchlist |
| GPT-5.5 standard, short context | $5 | $0.50 | $30 | $2.50 / $15 | OpenAI-native premium lane |
| GPT-5.5 standard, long context | $10 | $1 | $45 | $5 / $22.50 | Long context when OpenAI fit matters |
| GPT-5.5 Pro, short context | $30 | not listed | $180 | $15 / $90 | Highest-cost OpenAI Pro lane |
Cost Sanity Check
For a 100K-input / 20K-output request:
| Model | Approx standard cost |
|---|---|
| Claude Sonnet 5 | $0.40 introductory / $0.60 standard |
| Claude Opus 4.8 | $1.00 |
| GPT-5.5 standard short context | $1.10 |
| GPT-5.5 standard long context | $1.90 |
| Claude Fable 5 | $2.00 |
| GPT-5.5 Pro short context | $6.60 |
Those numbers are only a starting point. Repeated-context workloads can favor OpenAI or Claude cache-hit pricing. Batch workloads halve both providers’ listed input and output token charges. Data residency, priority/fast modes, and tool/runtime charges can change the bill.
When Claude Wins
Choose Claude when:
- Sonnet 5 solves the task and you want Claude behavior at the lowest current Claude production price.
- Opus 4.8 improves architecture, debugging, review, or long-horizon agent work enough to justify $5/$25.
- You need Claude Code or claude.ai as the first-party workflow.
- You want to test Fable only after target-route access is confirmed, Opus misses a high-value task, and your data policy allows Fable’s retention terms.
Do not route sensitive workloads to Fable just because a benchmark or launch quote looks strong. Fable can refuse or fall back on guarded domains, and Anthropic documents 30-day retention with no zero-data-retention option for Fable/Mythos.
When OpenAI Wins
Choose GPT-5.5 when:
- Your stack is already OpenAI-native or Codex-native.
- Cached input dominates the bill.
- The long-context standard lane is a better fit than Claude’s model behavior.
- You want GPT-5.5 standard pricing before considering much more expensive Pro routing.
GPT-5.5 standard short context is cheaper than Fable on both input and output. GPT-5.5 long context matches Fable input pricing and is cheaper on output. GPT-5.5 Pro is much more expensive than Fable, so use it only when the Pro behavior is measurably worth it.
Decision Matrix
| If your workload is… | Start with | Escalate to |
|---|---|---|
| Routine production coding | Claude Sonnet 5 or GPT-5.5 standard | Opus 4.8 if review/architecture quality matters |
| Premium Claude review | Opus 4.8 | GPT-5.5 or another live lane; Fable only after access, compliance, and guardrail tests pass |
| OpenAI-native coding | GPT-5.5 standard | GPT-5.5 Pro only after standard fails |
| Long-context repeated prompts | Compare cache-hit pricing on both providers | Batch or long-context lanes after a real bill estimate |
| Cyber, biology, chemistry, or model-development work | Guardrail test first | Trusted access or another approved model |
| Zero-data-retention requirement | Avoid Fable/Mythos-class routes | Use a model/contract that meets the requirement |
What To Test Before Buying
- Representative prompts from your real workload.
- Cost per successful task, not cost per token alone.
- Fallback/refusal logs for Fable-sensitive domains.
- Cache-hit rate on repeated context.
- Batch fit for asynchronous workloads.
- Data retention, data residency, and audit logging requirements.
- Tooling fit: Claude Code, Codex, cloud platform, or BYOK harness.
Sources
- Anthropic: Redeploying Fable 5 (Archive)
- Anthropic API docs: Claude Fable 5 and Claude Mythos 5 (Archive)
- Anthropic API docs: Pricing (Archive)
- Anthropic: Claude Sonnet 5 launch (Archive)
- Anthropic Platform: Sonnet 5 migration guide (Archive)
- OpenAI API docs: Pricing (Archive)
- OpenAI: GPT-5.6 limited preview (Archive)
- Anthropic: Claude Fable 5 and Claude Mythos 5 (Archive)
- Anthropic: Statement on the US government directive to suspend access to Fable 5 and Mythos 5 (Archive)
- WIRED: Anthropic Says It’s Taking Claude Fable 5 Offline to Comply With US Government Order (Archive)
- The Verge: Anthropic cuts off Fable 5 and Mythos 5 access following government order (Archive)
- WIRED: Anthropic Is Still at Odds With the White House Over Claude Fable 5 (Archive) — June 15: talks unresolved
- The Verge: Trump’s Anthropic shutdown just made the case for non-American AI (Archive) — June 15: sovereign AI reaction
- Business Insider: An AI startup is suing the US government for taking away Anthropic’s new model (Archive) — June 24: reported Fable restoration behind nationality and onboarding controls
Related links
- /compare/models/premium/ — Fable, Opus, Sonnet, and GPT-5.5 decision ladder
- /posts/fable-5-returns-claude-sonnet-5/ — July 1 Fable restoration and Sonnet 5 default-lane analysis
- /posts/claude-fable-5-mythos-5-cost-guardrails/ — Fable guardrails and buying test
- /value/smart-spend/ — Subscription and low-cost overflow strategy
- /models/claude-opus-4-5/ — Historical Opus 4.5 reference
Last updated: July 2, 2026. Fable 5 is restored on native Claude surfaces; Mythos 5 and GPT-5.6 remain restricted-access context. Pricing, included usage, cloud availability, retention, and model terms can change independently.