text-embedding-3-small
text-embedding-3-small is OpenAI's cheaper January 2024 embedding snapshot, still listed, not an unaudited 2026 crown. Rechecked 2026-08-18: leftover copy still called it a significant upgrade over ada-002 and a milestone in OpenAI's embedding line. openai.com/api/pricing still lists text-embedding-3-small at $0.020 / 1M tokens, beside text-embedding-3-large at $0.130 / 1M and leftover text-embedding-ada-002 at $0.100 / 1M. Drop leftover latest / milestone / 23% MTEB lift as a 2026 census.
Compare leftover text-embedding-3-large if you wanted the still-listed larger v3 row, leftover text-embedding-ada-002 if you wanted the older ada page, and leftover Qwen3-Embedding if you wanted a newer open embedding family.
Key Features
- Still a listed cheaper v3 row: First-party pricing still prints $0.020 / 1M.
- Not a 2026 crown: Do not keep leftover milestone / 23% MTEB copy as an audited 2026 score.
- Not ada-002: The older leftover page stays $0.100 / 1M.
- Website is the pricing page: Point at the live embeddings block, not platform API keys.
Use Cases
- People with a 3-small bookmark who need the leftover upgrade copy dropped.
- People comparing leftover 3-small vs leftover 3-large vs leftover ada-002.
- People who were about to paste a 23% MTEB lift into a 2026 deck.
Limitation: We did not re-run MTEB. Recheck first-party cards before you quote a 2026 score.
Pricing
First-party pages on 2026-08-18.
| Piece | Price | Notes |
|---|---|---|
| text-embedding-3-small | $0.020 / 1M | Still listed on pricing. |
| text-embedding-3-large | $0.130 / 1M | text-embedding-3-large. |
| text-embedding-ada-002 | $0.100 / 1M | text-embedding-ada-002. |
If a leftover blog still calls 3-small the current embedding milestone, treat the 2026 pricing block as the correction.
Getting Started
- Open openai.com/api/pricing and confirm the 3-small row.
- Recheck leftover text-embedding-3-large if you wanted the larger v3 twin.
- Recheck leftover text-embedding-ada-002 if production is still on ada.
- Do not paste leftover 23% MTEB into a 2026 deck.
Frequently Asked Questions
Is 3-small still listed?
Yes. First-party pricing still prints $0.020 / 1M.
Is it the 2026 embedding default?
Do not call it a crown. Recheck the live embeddings block before you pin a 2026 default.
Same as ada-002?
No. text-embedding-ada-002 is the older leftover at $0.100 / 1M.
Alternatives
- text-embedding-3-large: Still-listed larger v3 row.
- text-embedding-ada-002: Older leftover ada snapshot.
- Qwen3-Embedding: Newer open embedding family, also being rewritten as leftover.
Tips
- Call 3-small a still-listed cheaper v3 snapshot, not a 2026 crown.
- Quote $0.020 / 1M only while first-party still prints it.
- Recheck first-party cards before you invent an MTEB lift.
Conclusion
text-embedding-3-small is a leftover cheaper v3 snapshot still listed at $0.020 / 1M, not a 2026 embedding crown. Start at openai.com/api/pricing, then decide whether leftover text-embedding-3-large already covers the larger v3 path.
Comments
No comments yet. Be the first to comment!
Related Tools
text-embedding-3-large
developers.openai.com/api/docs/models/text-embedding-3-large
2024 OpenAI embedding snapshot. Rechecked: still listed. Not an unaudited 2026 flagship or MIRACL census.
NV-Embed-v2
huggingface.co/nvidia/NV-Embed-v2
2024 NVIDIA embedding snapshot. Rechecked: HF card still listed. Not a 2026 MTEB #1 census.
text-embedding-ada-002
openai.com/api/pricing
Previous OpenAI embedding snapshot. Rechecked: still listed at $0.100/1M. Not the 2026 embedding default.
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.