// COMPARISON

Ring-1T

A head-to-head benchmark comparison of Ring-1T across 25 evaluations.

The models

Language Open (Restricted)
Params
1T
Context
Released
Oct 2025

Today, we officially launch the trillion-parameter thinking model, Ring-1T. It is open-source upon release—developers can download the model weights from Huggin.

Full Ring-1T specs →

Benchmark comparison

Benchmark Ring-1T
code
Codeforces Elo 2,088
LiveCodeBench v6 78.3%
coding
Aider Benchmark 78.57
general
ArenaHard v2 81.59
C-Eval 91.53%
IFEval 85.21%
MMLU-Pro 80.54%
MultiChallenge 50.92%
knowledge
AGIEval 88.13
CMMLU 89.14
TriviaQA 78.59
math
AIME 2025 93.4%
CNMO 2024 88.54%
HMMT February 2025 86.72%
Omni-MATH 82.63
UGMathBench 76.47
reasoning
ARC-AGI-1 55.94
BBEH 59.63
FinanceReasoning 87.42
GPQA Diamond 78.63%
Humanity's Last Exam 16.03%
ZebraLogic 95.15%
safety
HealthBench 57.93%
science
PhyBench 42.65
tool-use
BFCL v3 68.82%

Best result per row highlighted in cyan. Each benchmark links to its definition and sources; each model links to its full scorecard.

Frequently asked questions

Which of these models is best for coding?

Ring-1T has the strongest result on Codeforces Elo among the models compared here. See the code-category rows in the table for the full picture.

How many benchmarks are compared?

This comparison covers 25 benchmarks on which at least one of the selected models has a published score.

Where do the benchmark scores come from?

Scores are aggregated from official model cards, technical reports and standard public evaluations, and link back to each benchmark's source. They are updated as new results are published.

Explore further

Build a different comparison

Models (0 selected)

Benchmarks (auto)

Select at least two models.