text-embedding-3-large
text-embedding-3-large is OpenAI's January 2024 embedding snapshot, still listed in the API docs, not an unverified 2026 flagship. The live model index on the public page still has an Embeddings section, but the public pricing HTML did not print a clean $ / 1M tokens row we could quote.
Compare text-embedding-ada-002 if you wanted the older ada page, Qwen3 Embedding if you wanted a newer open embedding family, and OpenAI if you wanted the live lab catalog.
Key Features
- Still a listed API object: 3-large remains a named embeddings model. Confirm the live card for current dimension and context fields.
- Not a 2026 crown: Do not keep MIRACL 54.9% or MTEB 64.6% as audited 2026 scores unless you re-run the first-party note.
- Matryoshka is leftover product copy: Older docs described shortening dimensions. Confirm the current card before you promise 256-3072 as law.
- No invented token dollar: Quote the live pricing page after you open a row, not this snapshot.
Use Cases
- People with a 2024 3-large integration who need the date attached.
- People who were about to paste MIRACL 54.9% into a 2026 deck.
Limitation: Pricing HTML in this pass did not yield a quoteable 3-large row. Confirm the official pricing table.
Pricing
OpenAI docs on 2026-08-17.
| Piece | Price | Notes |
|---|---|---|
| text-embedding-3-large | Confirm API pricing | Still listed. No invented $/1M row here. |
| Newer OpenAI embeddings | Confirm the model index | Do not assume 3-large is still the recommended default. |
If a leftover blog still calls 3-large the most advanced 2026 embedding, treat this page as the correction.
Getting Started
- Open the model page and the live pricing table.
- Confirm whether the current docs still recommend this SKU.
- Confirm Qwen3 Embedding if you wanted a newer family.
- Do not paste MIRACL 54.9% into a deck.
Frequently Asked Questions
Is this still OpenAI's best embedding?
Not something this page claims for 2026. Confirm the live catalog.
Are the 2024 MIRACL numbers still official?
We did not re-verify them. They stay off the table.
Same as ada-002?
No. ada-002 is the older leftover page.
Alternatives
- text-embedding-ada-002: Older OpenAI embedding snapshot.
- Qwen3 Embedding: Newer Qwen embedding family.
- OpenAI: Live lab and API catalog.
Tips
- Call 3-large a January 2024 snapshot that is still listed.
- Confirm the live recommended embedding before you write a contract.
Conclusion
text-embedding-3-large is a 2024 OpenAI embedding snapshot still in the docs, not an audited 2026 flagship. Start at the model page, then decide whether a newer embedding family already covers the job.
Comments
No comments yet. Be the first to comment!
Related Tools
Related Insights
Grok Bot and Hermes Bot: one person finally gets a think tank and a secretariat
Grok Bot now ships with Cursor Pro+. Hermes Bot runs on a VPS. They are not smarter chat boxes. The think tank advises, the secretariat executes, and you still make the call.
Hook OpenCode Go into Codex on Windows. Do Not Open a Second Toolkit.
A ChatGPT-signed Codex desktop app still shows mostly GPT in the picker. On Windows, enable only OpenCode Go and the Grok, GLM, Kimi, DeepSeek, and MiniMax models you already pay for appear in the same selector. Keys stay local. Native GPT stays put.
What Locks Codex Is the Picker, Not the Models
You already pay for OpenCode Go, Grok, and Z.ai, but the Codex picker still shows mostly GPT. The community project codex-router does not teach another install ritual. It puts subscriptions you already bought back into the selector, keeps keys on the machine, and leaves native GPT alone.