Skip to main content
BenchLM
Data
Recommendation
Ranking data as of

Best AI Models for Web Research in 2026

As of November 19, 2026, the top model among AI models for web research on the BenchLM leaderboard is Atria Dawn Preview with a score of 92.5.

Bottom line: Web research agents need to browse, gather evidence, and synthesize findings. BrowseComp is the most predictive benchmark here.

Full Rankings (64 ranked)

1
Atria Dawn Preview
Shanghai Artificial Intelligence Laboratory·Open Weight·256K

92.5

sourced avg

vs #2
2
GPT-5.6 Sol
OpenAI·Proprietary·1.05M

92.2

sourced avg

vs #3
3
Kimi K3.1 Pro
Moonshot AI·Open Weight·1.05M

92.1

sourced avg

vs #4
4
LongCat-2.5
Meituan·Open Weight·1M

91.9

sourced avg

vs #5
5
Ling 3.1 Flash
InclusionAI·Open Weight·262K

91.7

sourced avg

vs #6
6
GPT-6 Astra
OpenAI·Proprietary·1.05M

91.5

sourced avg

vs #7
7
Kimi K3
Moonshot AI·Open Weight·1.05M

91.2

sourced avg

vs #8
8
Claude Opus 5
Anthropic·Proprietary·

90.8

sourced avg

vs #9
9
Inkling 2 Small
Thinking Machines Lab·Proprietary·1M

90.2

sourced avg

vs #10
10
GPT-5.5 Pro
OpenAI·Proprietary·1M

90.1

sourced avg

vs #11
11
Fara1.5-27B
Microsoft·Open Weight·262K

89.3

sourced avg

vs #12
12
GPT-5.4 Pro
OpenAI·Proprietary·1.05M

89.3

sourced avg

vs #13
13
Step 5 Preview
StepFun·Open Weight·1M

88.7

sourced avg

vs #14
14
Claude Mythos 5
Anthropic·Proprietary·1M+

88

sourced avg

vs #15
15
GPT-5.6 Terra
OpenAI·Proprietary·1.05M

87.5

sourced avg

vs #16
16
ERNIE 5.5 Flash Preview
Baidu·Pending·1M

87.4

sourced avg

vs #17
17
Ornith-1.5-397B
Ornith AI·Open Weight·262K

86.6

sourced avg

vs #18
18
Claude Sonnet 5
Anthropic·Proprietary·1M

84.7

sourced avg

vs #19
19
Mistral Large 4
Mistral·Open Weight·1M

84.5

sourced avg

vs #20
20
GPT-5.5
OpenAI·Proprietary·1M

84.4

sourced avg

vs #21
21
Claude Opus 4.8
Anthropic·Proprietary·1M

84.3

sourced avg

vs #22
22
Claude Opus 4.6
Anthropic·Proprietary·1M

83.7

sourced avg

vs #23
23
MiniMax M3
MiniMax·Open Weight·1M

83.5

sourced avg

vs #24
24
DeepSeek V4 Pro
DeepSeek·Open Weight·1M

83.4

sourced avg

vs #25
25
dots3-note Preview
Dots Studio·Open Weight·512K

83.3

sourced avg

vs #26
26
GPT-5.6 Luna
OpenAI·Proprietary·1.05M

83.3

sourced avg

vs #27
27
Kimi K2.6
Moonshot AI·Open Weight·256K

83.2

sourced avg

vs #28
28
GPT-5.4
OpenAI·Proprietary·1.05M

82.7

sourced avg

vs #29
29
Sage-A1.5
InternScience·Open Weight·262K

81.7

sourced avg

vs #30
30
Fara1.5-4B
Microsoft·Open Weight·262K

80.8

sourced avg

vs #31
31
Ling 3.0 Flash Sante
InclusionAI·Open Weight·262K

80.2

sourced avg

vs #32
32
Claude Opus 4.7
Anthropic·Proprietary·1M

79.3

sourced avg

vs #33
33
Nemotron 3.5 Super VL 120B A12B
NVIDIA·Open Weight·1M

78

sourced avg

vs #34
34

77.5

sourced avg

vs #35
35
Inkling-Small
Thinking Machines Lab·Open Weight·1M

77.4

sourced avg

vs #36
36
Inkling
Thinking Machines Lab·Open Weight·1M

77.1

sourced avg

vs #37
37
Beam 1.0 501B-A23B
Reflection AI·Open Weight·1M

76.8

sourced avg

vs #38
38
Beam 1.0 501B-A23B FP8
Reflection AI·Open Weight·1M

76.7

sourced avg

vs #39
39
Step 3.7 Flash
StepFun·Open Weight·256K

75.8

sourced avg

vs #40
40
Beam 1.0 501B-A23B NVFP4
Reflection AI·Open Weight·1M

75.6

sourced avg

vs #41
41
Agents-A1
InternScience·Open Weight·262K

75.5

sourced avg

vs #42
42
DeepSeek V4 Flash 0731
DeepSeek·Open Weight·1M

73.2

sourced avg

vs #43
43
Ling 3.0 Flash
InclusionAI·Open Weight·262K

72.2

sourced avg

vs #44
44
Qwen3.5 397B
Alibaba·Open Weight·128K

69

sourced avg

vs #45
45
GLM-5.1
Z.AI·Open Weight·203K

68

sourced avg

vs #46
46
Ornith-1.5-35B-A3B
Ornith AI·Open Weight·262K

67.6

sourced avg

vs #47
47
Agents-A1-4B
InternScience·Open Weight·262K

66.8

sourced avg

vs #48
48
GPT-5.2
OpenAI·Proprietary·400K

65.8

sourced avg

vs #49
49
Qwen3.5-122B-A10B
Alibaba·Open Weight·262K

63.8

sourced avg

vs #50
50
Nemotron 3.5 Bolt 74B-A9B BF16
NVIDIA·Open Weight·256K

62.1

sourced avg

vs #51
51
GLM-5
Z.AI·Open Weight·200K

62

sourced avg

vs #52
52
Qwen3.5-27B
Alibaba·Open Weight·262K

61

sourced avg

vs #53
53
Qwen3.5-35B-A3B
Alibaba·Open Weight·262K

61

sourced avg

vs #54
54
Kimi K2.5
Moonshot AI·Open Weight·256K

60.6

sourced avg

vs #55
55
Nemotron 3.5 Bolt 74B-A9B
NVIDIA·Open Weight·256K

60.4

sourced avg

vs #56
56
Ornith-1.5-9B
Ornith AI·Open Weight·262K

56.4

sourced avg

vs #57
57
Spark-X2.6-4B
SparkLLM·Open Weight·1M

53.1

sourced avg

vs #58
58
GLM-4.7
Z.AI·Open Weight·200K

52

sourced avg

vs #59
59
Sarvam 105B
Sarvam·Open Weight·128K

49.5

sourced avg

vs #60
60
Solar Pro 4
Upstage·Proprietary·512K

49.2

sourced avg

vs #61
61
LongCat-Flash-Lite-Sparse
Meituan·Open Weight·1M

48.6

sourced avg

vs #62
62
Nemotron 3 Ultra
NVIDIA·Open Weight·1M

44.4

sourced avg

vs #63
63

36.8

sourced avg

vs #64
64
Sarvam 30B
Sarvam·Open Weight·64K

35.5

sourced avg

How to choose

Key Takeaways

The top model on this sourced reporting-family slice is Atria Dawn Preview by Shanghai Artificial Intelligence Laboratory with an average of 92.5.

The best open-weight model is Atria Dawn Preview at position #1.

64 models are listed with sourced benchmark coverage in this reporting family.

Score in Context

What these scores mean

This ranking averages sourced web-research benchmarks. It isolates the browsing and evidence-gathering slice of agentic performance.

Known limitations

Web research benchmarks test specific browsing patterns. Real-world web research also depends on access to search APIs, page rendering quality, and anti-bot measures that benchmarks do not capture.

About this ranking

Ranking data as of November 19, 2026

This reporting page isolates the web research slice of agentic performance. It prioritizes sourced benchmarks for browsing, evidence gathering, and multi-step web task completion rather than generic overall agent scores.

This page ranks models using only sourced web research benchmarks in the reporting family.

Atria Dawn Preview leads this ranking with a score of 92.5, followed by GPT-5.6 Sol (92.2) and Kimi K3.1 Pro (92.1). The top three are separated by just a few points — any of them would perform well for this use case.

The best open-weight option is Atria Dawn Preview (ranked #1 with a score of 92.5). Open-weight models are highly competitive in this category — self-hosting is a viable alternative to proprietary APIs.

This ranking uses provisional overall weighted scores from the active scoring formula. For detailed model profiles, click any model name above. To compare two specific models head-to-head, use the "vs #" links.

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 7,000+ subscribers.

One email each week. Unsubscribe anytime.