The AI arena is free today

Open Superagent

Beam 1.0 vs MAI-Thinking-1

Beam 1.0 leads the LLM Stats Score 44.6 to 32.6.

Reflection AI · Microsoft · Updated for 2026

Which is better?

Beam 1.0 leads the overall LLM Stats Score 44.6 to 32.6, ranking #46 overall.

In the 6 individual benchmarks reported for both models, Beam 1.0 wins 6; this is a narrower head-to-head signal than the composite indexes.

Based on current LLM Stats indexes, shared benchmarks, pricing, and model metadata for 2026.

Choose Beam 1.0

  • overall performance matters — it scores 44.6 and ranks #46 on LLM Stats
  • your work emphasizes reasoning and coding — it leads those capability indexes
  • you value its reported benchmark strengths — it wins 6 of 6 exact shared results
  • you want the most recent training data — it shipped Oct 2026
  • you need open weights you can self-host or fine-tune

Choose MAI-Thinking-1

  • you are already invested in the Microsoft ecosystem

At a glance

The differences that matter most.

Beam 1.0
Reflection AI
MAI-Thinking-1
Microsoft
Core performance indexes
LLM Stats Score
44.6
#46
32.6
#124
Reasoning
45.5
#37
33.5
#114
Coding
33.7
#39
19.2
#123
Agents
28.8
#53
12.8
#120
Cost, coverage & limits
Benchmark wins
6 of 6
0 of 6
Input price
$0.40 / M
— / M
Output price
$1.20 / M
— / M
Context window
1,048,576
—

Capability indexes

Additional strengths measured across groups of related public benchmarks

3 shared
Index
Beam 1.0
MAI-Thinking-1
Math
35.1#45
33.6#55
Tool use
26.4#32
10.6#139
Long context
29.5#5
18.8#39
Conservative TrueSkill rating · higher is betterHow scores work

Individual benchmarks

22 reported for Beam 1.0 · 23 for MAI-Thinking-1

6 shared

Beam 1.0 outperforms in 6 benchmarks (AIME 2026, GPQA, IFBench, LongBench v2, SWE-Bench Pro, SWE-Bench Verified), while MAI-Thinking-1 is better at 0 benchmarks.

Beam 1.0 significantly outperforms across most benchmarks.

Thu Oct 22 2026 • llm-stats.com

Human preference

Blind head-to-head votes and playground preference scores

Model Size

Parameter count comparison

498.7B diff

MAI-Thinking-1 has 498.7B more parameters than Beam 1.0, making it 99.5% larger.

Reflection AI
Beam 1.0
501.3Bparameters
Microsoft
MAI-Thinking-1
1.0Tparameters
501.3B
Beam 1.0
1000.0B
MAI-Thinking-1

Context Window

Maximum input and output token capacity

Only Beam 1.0 specifies input context (1,048,576 tokens). Only Beam 1.0 specifies output context (131,072 tokens).

Reflection AI
Beam 1.0
Input1,048,576 tokens
Output131,072 tokens
Microsoft
MAI-Thinking-1
Input- tokens
Output- tokens
Thu Oct 22 2026 • llm-stats.com

License

Usage and distribution terms

Beam 1.0 is licensed under Apache 2.0, while MAI-Thinking-1 uses a proprietary license.

License differences may affect how you can use these models in commercial or open-source projects.

Beam 1.0

Apache 2.0

Open weights

MAI-Thinking-1

Proprietary

Closed source

Release Timeline

When each model was launched

Beam 1.0 was released on 2026-10-16, while MAI-Thinking-1 was released on 2026-06-02.

Beam 1.0 is 5 months newer than MAI-Thinking-1.

Beam 1.0

Oct 16, 2026

4 days ago

4mo newer
MAI-Thinking-1

Jun 2, 2026

4 months ago

Knowledge Cutoff

When training data ends

Beam 1.0 has a documented knowledge cutoff of 2026-06-30, while MAI-Thinking-1's cutoff date is not specified.

We can confirm Beam 1.0's training data extends to 2026-06-30, but cannot make a direct comparison without MAI-Thinking-1's cutoff date.

Beam 1.0

Jun 2026

MAI-Thinking-1

—

Outputs Comparison

Notice missing or incorrect data?

Judge for yourself.

Run your own prompts against Beam 1.0 and MAI-Thinking-1 side-by-side, then vote on the output you prefer.

Beam 1.0
✓ Preferred
MAI-Thinking-1
Open in Playground

FAQ

Common questions about Beam 1.0 vs MAI-Thinking-1.

Which is better, Beam 1.0 or MAI-Thinking-1?

Beam 1.0 leads the LLM Stats Score 44.6 to 32.6. Beam 1.0 is made by Reflection AI and MAI-Thinking-1 is made by Microsoft. The best choice depends on your use case — compare their capability indexes, individual benchmarks, pricing, and limits above.

How does Beam 1.0 compare to MAI-Thinking-1 in benchmarks?

Beam 1.0 scores AIME 2026: 97.7%, GPQA: 90.6%, LiveCodeBench: 88.3%, SWE-Bench Verified: 81.6%, MCP Atlas: 80.8%. MAI-Thinking-1 scores LongFact: 98.0%, AIME 2025: 97.0%, AIME 2026: 94.5%, GraphWalks: 90.0%, AIR-Bench: 88.0%.

What are the context window sizes for Beam 1.0 and MAI-Thinking-1?

Beam 1.0 supports 1.0M tokens and MAI-Thinking-1 supports an unknown number of tokens. A larger context window lets you process longer documents, conversations, or codebases in a single request.

What are the main differences between Beam 1.0 and MAI-Thinking-1?

Key differences include LLM Stats Score (44.6 vs 32.6), licensing (Apache 2.0 vs Proprietary). See the full comparison above for benchmark-by-benchmark results.

Who makes Beam 1.0 and MAI-Thinking-1?

Beam 1.0 is developed by Reflection AI and MAI-Thinking-1 is developed by Microsoft.

Related comparisons