whatmodel.sh / Ring-2.6-1T
Ring-2.6-1T
↗ share
+ contribute
Ring-2.6-1T is an open-weights language model from inclusionAI (Ant Group), released May 2026. Its published size is 1025B parameters (63B active per token). Licence: mit.
Catalogue identity notes
Comparison-only. inclusionAI's Ring-2.6-1T reasoning model (repo inclusionAI/Ring-2.6-1T @2314abd, created 2026-05-14, MIT; card: 'a trillion-parameter flagship reasoning model', reasoning effort levels high and xhigh). Added because the same lab's Ling-3.0-flash card prints a 'Ring-2.6-1T (xhigh)' column on 14 benchmarks that otherwise hold only Ling-3.0-flash. No verified run route recorded.
Model information sources
Release date ↗ · Parameter count ↗
1,025,414,425,344 parameters in the repo's safetensors metadata; the Ling-3.0-flash card's table prints the size as '1T-A63B' (63B active).
$ how it compares
The LLMs ranked either side of Ring-2.6-1T, out of 828 with a score, then more from inclusionAI (Ant Group).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 98 |  Nemotron 3 Ultra 550B-A55B | 69±1 | api $1.35/M +local | ||
| 99 |  Qwen3.6 27B | 69±1 | api $1.20/M +local | ||
| 100 | Ring-2.6-1T | 69±2 | comparison only | ||
| 101 |  A.X K2 (688B-A33B) | 69±1 | local | ||
| 102 |  o3 | 69±1 | api $3.50/M | ||
| more from inclusionAI (Ant Group) | |||||
| 78 | Ling 3.0 Flash (124B-A5B) | 71±2 | api $0.090/M +local | ||
$ what it's good at
agent tasks█··········█ #2 of 2 · 25% of its score · 3 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SWE-bench Pro (split unspecified) lab | 53.9% ✎ | ▃·▁▁▁▂▂▃█▇▃▁ beats 29/55 | 14% |
| SWE-bench Multilingual lab | 56.7% ✎ | ▁··▁▃▂▅▇█▃▃▁ beats 17/60 | 7.2% |
| BFCL V4 (December 2025 archive) lab | 64.8% ✎ | ▁▁▁▂▇▆▆▆█▂▄▂ beats 96/111 | 4.5% |
coding█··········█ #2 of 2 · 27% of its score · 3 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SWE-bench Pro (split unspecified) lab | 53.9% ✎ | ▃·▁▁▁▂▂▃█▇▃▁ beats 29/55 | 14% |
| SWE-bench Multilingual lab | 56.7% ✎ | ▁··▁▃▂▅▇█▃▃▁ beats 17/60 | 7.2% |
| LiveCodeBench 2024-08–2025-05 (version unstated) lab | 87.0% ✎ | ▃···█··▅··▃▃ beats 7/7 | 5.9% |
math█··········█ #2 of 2 · 33% of its score · 3 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| AIME 2026 lab | 95.8% ✎ | ▁▁▁▂▁▂▆█▆▂·▁ beats 44/49 | 12% |
| IMO-AnswerBench lab | 86.1% ✎ | ▁▁·▁▁▁▄▅█▄▂▃ beats 34/42 | 11% |
| HMMT February 2026 lab | 93.5% ✎ | ▃▃▂▃·██▆█▅▃▅ beats 31/33 | 10% |
other tests7 not in a task · 35% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Humanity's Last Exam lab | 18.3% ✎ | ▁▁▁▃▆▆▇██▆▁▃ beats 46/104 | 8.9% |
| SkillsBench 87-task subset lab | 11.9% ✎ | ▄··▄▄▄···█·█ beats 0/7 | 7.1% |
| WideSearch corrected dataset lab | 62.2% ✎ | ▃······▃▃▃█▃ beats 2/7 | 7.1% |
| SysBench (aggregation unspecified) lab | 86.5% ✎ | █··█····█··█ beats 2/3 | 4.6% |
| Multi-IF (aggregation unspecified) lab | 89.3% ✎ | ▁▁▁▁▁▁▃▄█▄▁▃ beats 41/41 | 4.0% |
| IFBench (aggregation unspecified) lab | 44.6% ✎ | ▃▇▂▄▅▅▂▄▄▆█▇ beats 31/80 | 3.2% |
| MCP-Atlas v1 public500 lab | 61.2% ✎ | █··········█ beats 0/1 | — |
$ where to run it
No verified API or download recorded yet · add one
Official weights and documentation: huggingface.co/inclusionAI ↗
$ how the score was worked out · inclusionAI/Ling-3.0-flash mo… 99%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 inclusionAI/Ling-3.0-flash model card · lab-reported | 13 | 99% |
| prior · pull toward the middle | 1% | |
| weighted average | = 69.4 |
not counted (1)
| inclusionAI/Ling-3.0-flash model card | MCP-Atlas v1 public500 [correct] | 61.2% | fewer than three models have this test yet, or this source ran it for this model alone |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json