Comparison · verified August 21, 2026
Claude vs OpenAI for Enterprise
Enterprise selection is not a model comparison. By the time a deal reaches procurement, both vendors clear the capability bar and the decision turns on governance surfaces, auditability, deployment flexibility, and what happens when something goes wrong.
Anthropic has shipped harder on the governance surface through 2026 — inference hooks, Compliance API coverage down to local sessions, plugin scanning, jurisdiction pinning. If your security review has previously rejected an AI vendor, that is the material difference. If your organisation runs on Azure, OpenAI’s integration depth may still outweigh it.
The verified differences
What each one costs
Last verified August 21, 2026
USD per million tokens, standard rates, no discounts applied. Read the three caveats underneath before you put these numbers in a spreadsheet.
| Model | Tier | Input | Output | Notes |
|---|---|---|---|---|
| Claude Fable 5 | frontier | $10 | $50 | The frontier model. Twice the price of Opus 5 for the hardest work. |
| Claude Opus 5 | frontier | $5 | $25 | The default flagship. Thinking on by default. Fast mode available at $10/$50. |
| Claude Sonnet 5 | balanced | $2 | $10 | What most production agent fleets run on. The September 2026 increase to $3/$15 was cancelled. |
| Claude Haiku 4.5 | fast | $1 | $5 | High-volume and latency-sensitive work. |
| GPT-5.6 Sol | frontier | $5 | $30 | The flagship tier. Long-context requests meter at $10/$45. |
| GPT-5.6 Terra | balanced | $2 | $12 | The everyday tier, cut to this rate on July 30, 2026. Long-context meters at $4/$18. |
| GPT-5.6 Luna | fast | $0.20 | $1.2 | The cheapest frontier-family option on the market. Long-context meters at $0.40/$1.80. |
Anthropic changed tokenizers at Claude 4.7
Claude 4.7 and later — including Opus 5, Sonnet 5, and Fable 5 — use a newer tokenizer that produces roughly 30% more tokens for the same text than Sonnet 4.6 and earlier. A $2/MTok model on the new tokenizer is not directly comparable to a $2/MTok model on an older one, or to another vendor. Compare cost per task, not cost per token.
OpenAI meters long context separately
GPT-5.6 publishes a second, higher rate for long-context requests: Sol goes from $5/$30 to $10/$45, Terra from $2/$12 to $4/$18, Luna from $0.20/$1.20 to $0.40/$1.80. Anthropic includes the full 1M-token window at standard pricing on Claude 4.6 and later — a 900k-token request costs the same per token as a 9k one. If your workload is context-heavy, that difference is larger than the headline gap.
Two of Google’s current rates are introductory
Gemini 3.7 Flash and 3.6 Flash are priced at $0.75/$3.75 only through December 31, 2026. Both double to $1.50/$7.50 in 2027. If you are building a twelve-month cost model on Gemini Flash, model the 2027 number, not the one on the page today.
Caching and batching move the number more than model choice
A Claude cache hit costs 10% of the standard input price, and the Batch API takes 50% off both directions; the two stack. A workload that reuses a large system prompt can land below a nominally cheaper model that you are calling uncached. Model selection is usually the third-largest lever, after caching and after not sending the context at all.
What the evidence says, and who is saying it
Almost every performance number in public circulation traces back to a vendor running its own harness. That does not make the numbers useless, but the attribution belongs next to the claim.
Verifiable from release notes and vendor documentation
- Anthropic shipped inference hooks (Aug 5), plugin security scanning (Aug 6), Compliance API coverage of local Cowork and Claude Code sessions (Aug 11), and Admin API user management GA (Aug 19) — all in 2026.
- Claude models run on the Claude API, Bedrock, Claude Platform on AWS, Google Cloud, and Microsoft Foundry.
- Three Claude models were retired or suspended in the first half of 2026.
The shipping cadence is a fact; whether it maps to your specific control requirements is not. Take this list to your security team as a starting question set, not as an answer.
What we deliberately are not claiming
- That one vendor is more secure than the other.
- That either vendor’s certifications satisfy your regulator.
- That published controls behave as documented under your configuration.
Nobody outside your organisation can make those claims responsibly, and any comparison page that does is guessing. Security posture is a function of your configuration and your threat model, not the vendor’s feature list.
The procurement questions that actually settle it
Run these before any technical evaluation. Two of them have ended vendor selections at organisations we have worked with, and both were answerable in week one rather than month four.
- Can you block a prompt before the model sees it? If your DLP requirements need inline prevention rather than after-the-fact detection, ask both vendors this directly and get the answer in writing. It is the sharpest current differentiator.
- Can you retrieve a transcript of a session that ran on an employee’s laptop? Local sessions are the blind spot in most AI audit stories. Ask specifically about desktop and CLI sessions, not just the web app.
- What is the hard spend cap, and who can raise it? An alert is not a cap. Establish whether an unattended agent can run up an unbounded bill and what stops it.
- Where does inference physically run, and can you pin it? Get the mechanism and the price multiplier, not a general assurance about regions.
- What is the deprecation notice period, contractually? Both vendors retire models roughly annually. The question is not whether it will happen but how much warning you get and what your migration path costs. This belongs in the contract, not in a blog post.
- Run one real workload end to end, with security in the room. Not a demo. A workload with your actual data, your actual permissions, and the people who will have to sign off watching it. Most objections surface in the first hour of this and never in a slide deck.
Getting IT approval covers the internal path, the Compliance API covers the audit surface in detail, and when your model disappears covers the deprecation risk you are underwriting either way.
The honest bottom line
Pick Claude if governance is the binding constraint. Inference hooks, Compliance API coverage reaching local sessions, plugin scanning, jurisdiction pinning, and hard session budgets are a more developed control surface than the alternative in 2026, and the five deployment surfaces mean you are not underwriting a single cloud relationship.
Pick OpenAI if your organisation is deeply committed to Azure and the integration depth is worth more than the control-surface difference, or if your teams are already productive on the tooling and the switching cost is real. Familiarity is a legitimate procurement input, not a weak one.
Plan for both, whichever you sign Both vendors retired models in 2026, and one had two models suspended worldwide by government order for nineteen days. An abstraction layer over your model calls is a few days of work and it is the cheapest insurance available against a decision neither you nor your vendor controls.