Forecast: Pareto 26.10 vs 26.9 context and price on OpenRouter
26.10 is already listed: longer context and a lower per-million price. Quality, official weights, and a next date are absent.
On 30 October 2026 the OpenRouter catalog already lists Pareto 26.10 beside 26.9. This forecast only repeats the context and prices that listing actually returns.
Both blurbs call the model a multimodal composite for research, coding, and agent work: text and image in, text out. That is catalog copy, not a score. Pareto 26.10 versus 26.9 differs in fields, not in adjectives.
Figures the catalog already shows
OpenRouter prices are USD per token. The table multiplies by 1,000,000. Context on 26.10 is four times 26.9 (1048576 ÷ 262144).
| Field | Pareto 26.10 | Pareto 26.9 |
|---|---|---|
| Model id | unbiased/pareto-26.10 | unbiased/pareto-26.9 |
| Context | 1048576 | 262144 |
| Input (USD / 1M tokens) | 0.80 | 2.50 |
| Output (USD / 1M tokens) | 3.20 | 7.50 |
| Cache read (USD / 1M tokens) | 0.03 | 0.25 |
| created (UTC) | 2026-10-15 | 2026-10-13 |
| Modality | text+image → text | text+image → text |
| max_completion_tokens | 131072 | 131072 |
| hugging_face_id | empty | empty |
- 26.10 per-token rates
- prompt 0.0000008, completion 0.0000032, input_cache_read 0.00000003
- 26.9 per-token rates
- prompt 0.0000025, completion 0.0000075, input_cache_read 0.00000025
The newer row is already the alias target
Alias ~unbiased/pareto-latest has alias_target name Pareto 26.10 and slug unbiased/pareto-26.10. That alias row lists the same context and per-million prices as 26.10. A separate id, openrouter/pareto-code, is a coding router whose prompt and completion fields are -1. That is not a fixed quote for these two rows.
What the listing and the search do not show
No quality score
The model object has no eval score, latency, or knowledge cutoff. knowledge_cutoff and expiration_date are empty. A lower unit price does not mean better answers.
No official weights, no next date
A Hugging Face search for Pareto-26 on 30 October 2026 returned only rd211/Qwen2.5-7B-Instruct-Pareto0.5-260, created 2025-04-29, a Qwen2 repo. No official Pareto 26.10 weights were in that search. hugging_face_id is empty on both rows. No later Pareto 26 row is listed, so do not invent a release date.
Read the two rows in this order
- 01Compare context and unit price
For long context or token billing, use 1048576 versus 262144, and 0.80/3.20 versus 2.50/7.50 USD per million tokens.
- 02See where the alias points
On this fetch, latest already points at 26.10. Do not treat 26.9 as the current alias target.
- 03Leave blank fields blank
There is no quality score, no official weight repo, and no next release date. The two-day created gap does not forecast another version.
Price and context are checkable. These two rows do not show which answers are better, or on which day a later version appears.
After the numbers, a browser room can hold the comparison on one page. See the guide.
Is Pareto 26.10 already listed on OpenRouter?
Yes. The 30 October 2026 model list includes unbiased/pareto-26.10, created 2026-10-15 UTC. unbiased/pareto-26.9 is still listed, created 2026-10-13 UTC.
Are official Pareto 26.10 weights on Hugging Face?
Not in this search. The only Pareto-26 hit was the older Qwen repo rd211/Qwen2.5-7B-Instruct-Pareto0.5-260. hugging_face_id is empty on both OpenRouter rows.
How do context and price differ?
Context is 1048576 versus 262144. Per million tokens, 26.10 is about 0.80 USD in and 3.20 USD out; 26.9 is about 2.50 USD in and 7.50 USD out. Cache reads are about 0.03 versus 0.25 USD.
When does the next Pareto ship?
The catalog does not give a later release date. Do not treat created or the date suffix inside a canonical slug as a general-availability announcement.