Agentic work
Tool use, computer use, and multi-step task completion
GLM-5.2
GLM-5.2 has the higher public agentic point estimate, 61.9 to 55.2, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Updated November 19, 2026. Rank says GLM-5.2 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.
GLM-5.2 has the higher public point estimate, 59.2 versus 47.3. Their conditional score ranges overlap. These ranges do not establish rank confidence. 9 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Reflection AI
47.3/100
Supported · Public rank #104
Conditional range 34.7–59.8
GLM-5.2: $1.40 input / $4.40 output
Beam 1.0 501B-A23B: No comparable API price
GLM-5.2: 1M
Beam 1.0 501B-A23B: 1M
GLM-5.2: June 16, 2026
Beam 1.0 501B-A23B: October 16, 2026
Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.
Tool use, computer use, and multi-step task completion
GLM-5.2
GLM-5.2 has the higher public agentic point estimate, 61.9 to 55.2, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Code generation, repair, and software-engineering tasks
Not enough matched evidence
Beam 1.0 501B-A23B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like.
Prompts that approach the documented context limit
No clear pick
The documented context windows are equal.
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.
The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.
Directional only · BenchAlign v6.0
GLM-5.2 has the higher coding point estimate. Conditional score ranges do not establish rank confidence.
Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.
The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.
We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy
Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.
Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.
HLE w/o toolsKnowledge
Normalized gap 4.5SWE-bench ProCoding
Normalized gap 4.0AIME26Math
Normalized gap 1.5Each row shows the public-lane category score for both models: the BenchAlign v6.0 lane in every category, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence. Directional and not-comparable rows remain visible, but they never receive a winner.
| Category | GLM-5.2 | Beam 1.0 501B-A23B | Basis | Reading |
|---|---|---|---|---|
| Agentic | 61.9Supported · #51/177 | 55.2Supported · #76/177 | Like-for-likeBenchAlign v6.0 lane · 8 vs 7 public rows | GLM-5.2 leads · intervals overlap |
| Coding | 61.0Supported · #47/177 | 39.2Estimated · #139/177 | Directional onlyBenchAlign v6.0 lane · 11 vs 8 public rows | Directional only |
| Instruction following | 55.1Supported · #49/142 | 44.8Estimated · #95/142 | Directional onlyBenchAlign v6.0 lane · 0 vs 1 public rows | Directional only |
| Reasoning | 52.5Supported · #67/160 | Not ranked | Not comparableBenchAlign v6.0 lane · 1 vs 2 public rows | Not comparable |
| Multimodal | 60.1Estimated · #53/115 | Not ranked | Not comparableBenchAlign v6.0 lane · 0 vs 0 public rows | Not comparable |
| Knowledge | 55.9Estimated · #60/156 | Not ranked | Not comparableBenchAlign v6.0 lane · 6 vs 2 public rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableBenchAlign v6.0 lane · 0 vs 0 public rows | Not comparable |
| Math | 57.3Supported · #43/124 | Not ranked | Not comparableBenchAlign v6.0 lane · 6 vs 1 public rows | Not comparable |
Ranks count the models scored in each category’s BenchAlign v6.0 lane, so denominators differ by category. Like-for-like rows within 0.5 points read as a practical tie.
Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.
1K fresh input + 500 output tokens
Beam 1.0 501B-A23B has no comparable published API token rate.
50K fresh input + 3K output tokens
Beam 1.0 501B-A23B has no comparable published API token rate.
200K cached + 20K fresh input + 10K output tokens
GLM-5.2 has no published cached-input rate, so cached tokens use its listed input rate. Beam 1.0 501B-A23B has no comparable published API token rate.
Overall score against output price per 1M tokens, one model per family. Pricing as of November 18, 2026.
GLM-5.2
Gray dots are the other priced models. The dark step line is the efficiency frontier: for a model under it, another model costs no more and scores at least as high.
See GLM-5.2 on the full price vs performance chartSourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.
Maximum documented context; output-token limits may be lower.
GLM-5.2
1M
Beam 1.0 501B-A23B
GLM-5.2
Not sourced
Beam 1.0 501B-A23B
Beam-1.0-501B-A23B
Reflection developer docs, ModelsA missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
GLM-5.2
Not published
Beam 1.0 501B-A23B
No comparable hosted API rate
Reflection launch postGLM-5.2
Beam 1.0 501B-A23B
GLM-5.2
Beam 1.0 501B-A23B
GLM-5.2
Not sourced
Beam 1.0 501B-A23B
Open Weights · Hugging Face, self-hosted inference, Reflection API public beta, partner endpoints
Reflection Beam 1.0 BF16 model cardGLM-5.2
Reasoning
Beam 1.0 501B-A23B
Reasoning
GLM-5.2
Open Weight
Beam 1.0 501B-A23B
Open Weight
GLM-5.2
Open Weight
Beam 1.0 501B-A23B
Open Weight
GLM-5.2
2026-06-16
Beam 1.0 501B-A23B
2026-10-16
GLM-5.2 has the higher public point estimate, 59.2 versus 47.3. Their conditional score ranges overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.
GLM-5.2 scores higher for coding on the public lane, 61 to 39.2. Beam 1.0 501B-A23B is scored on Estimated evidence for coding, so the reading is directional rather than like-for-like. BenchLM does not name a winner for coding from a directional reading; read the shared benchmark rows directly and test the models on the same task set.
GLM-5.2 has the higher public agentic tasks point estimate, 61.9 to 55.2, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Both models do not have comparable published API token rates, so this page does not name a universal price winner.
Both models list the same context window, 1M.
Estimates at 50,000 req/day · 1000 tokens/req average.
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Terminal-Bench 3.0
Not directly comparable
Terminal-Bench 2.1
GLM-5.2 leads this result
MCP Atlas
Beam 1.0 501B-A23B leads this result
Toolathlon
Not directly comparable
ResearchClawBench
Not directly comparable
Terminal-Bench 2.1 (Vals)
Not directly comparable
HLE w/ tools
GLM-5.2 leads this result
MazeBench
Not directly comparable
Terminal-Bench 4.1
Not directly comparable
τ³-bench results
Not directly comparable
BrowseComp
Not directly comparable
DeepSearchQA
Not directly comparable
SWE-bench Pro
Beam 1.0 501B-A23B leads this result
NL2Repo
Not directly comparable
Terminal-Bench 2.1
GLM-5.2 leads this result
ProgramBench
Not directly comparable
CursorBench 3.2
Not directly comparable
OpenHarmony Bench v1.0
Not directly comparable
LiveCodeBench (Vals)
Not directly comparable
SWE-bench (Vals)
Not directly comparable
PostTrainBench v1.1
Not directly comparable
PostTrainBench
Not directly comparable
PostTrainBench v1
Not directly comparable
DeepSWE
Not directly comparable
SWE-Atlas Codebase QnA
Not directly comparable
SWE Multilingual
Not directly comparable
SWE-bench Verified
Not directly comparable
SciCode
Not directly comparable
VulcanBench Frontier v4
Not directly comparable
GPQA
Not directly comparable
GPQA-D
GLM-5.2 leads this result
HLE
Not directly comparable
HLE w/o tools
GLM-5.2 leads this result
GPQA Diamond (Vals)
Not directly comparable
MMLU-Pro (Vals)
Not directly comparable
IFBench
Not directly comparable
AIME26
GLM-5.2 leads this result
HMMT Nov 2025
Not directly comparable
HMMT Feb 2026
Not directly comparable
MMAnswerBench
Not directly comparable
FrontierMath v2 (Tiers 1-3)
Not directly comparable
FrontierMath v2 (Tier 4)
Not directly comparable
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 7,000+ subscribers.
Last updated November 19, 2026