GLM 5.2
GLM 5.2 is served by 34 companies at prices ranging from $0.06 to $2.50 per million input tokens, a 41.7× difference for identical weights.
- cheapest host · Wafer
- $0.06
- price spread across hosts
- 42×
- companies serving it
- 34
- intelligence index
- 33.7
Is it worth it?
Beaten on both axes. GLM 5.3 Flash scores 41.8 against 33.7 and costs 8.7× less for the same work.
Every host, compared
The same weights, priced and configured differently by 34 companies. Precision matters: 4-bit builds measurably degrade output and cost the same.
| Host | $/1M in | $/1M out | Cached | Context | Precision | Uptime |
|---|---|---|---|---|---|---|
| Wafercheapest | $0.060 | $7.00 | $0.059 | 1.05M | unknown | 100% |
| Wafer | $0.190 | $8.00 | $0.180 | 1.05M | unknown | 100% |
| Cloudflare | $0.190 | $8.00 | $0.190 | 262K | unknown | 100% |
| InferenceNet | $0.198 | $2.40 | $0.132 | 1.05M | fp4 | 100% |
| Decart | $0.260 | $1.60 | $0.100 | 1.05M | mxfp4 | 96.4% |
| Morph | $0.400 | $6.00 | $0.137 | 1.05M | fp8 | 99.9% |
| DeepInfra | $0.563 | $1.80 | $0.105 | 1.05M | fp4 | 99.9% |
| StreamLake | $0.643 | $2.02 | $0.119 | 1.02M | fp8 | 99.2% |
| Novita | $0.650 | $2.04 | $0.121 | 1.05M | fp8 | 100% |
| DigitalOcean | $0.700 | $2.20 | $0.105 | 1.05M | unknown | 100% |
| CoreWeave | $0.760 | $2.42 | $0.140 | 1.05M | fp4 | 100% |
| AtlasCloud | $0.938 | $2.95 | $0.174 | 1.05M | fp8 | 100% |
| Alibaba | $0.966 | $3.04 | $0.193 | 1.00M | fp8 | 99.7% |
| Alibaba | $1.12 | $3.52 | $0.224 | 1.00M | fp8 | 98.6% |
| SiliconFlow | $1.19 | $3.74 | $0.221 | 1.05M | fp8 | 100% |
| Inceptron | $1.25 | $5.20 | $0.200 | 1.05M | fp4 | 99.8% |
| Phala | $1.26 | $3.00 | $0.220 | 1.05M | fp8 | 100% |
| Nebius | $1.40 | $4.40 | n/a | 1.05M | fp4 | 98.9% |
| Mistral | $1.40 | $4.40 | $0.140 | 1.05M | nvfp4 | 100% |
| BaseTen | $1.40 | $4.40 | $0.140 | 1.05M | fp8 | 100% |
| Together | $1.40 | $4.40 | $0.260 | 1.05M | unknown | 100% |
| Baidu | $1.40 | $4.40 | $0.260 | 1.05M | fp8 | 99.7% |
| BaseTen | $1.40 | $4.40 | $0.140 | 1.05M | fp8 | 100% |
| Venice | $1.40 | $4.40 | $0.260 | 1.00M | fp8 | 99.9% |
| GMICloud | $1.40 | $4.40 | $0.260 | 1.05M | fp8 | 100% |
| Parasail | $1.40 | $4.40 | $0.260 | 262K | fp4 | 100% |
| Friendli | $1.40 | $4.40 | $0.260 | 1.05M | unknown | 99.9% |
| Z.AI | $1.40 | $4.40 | $0.260 | 1.05M | fp8 | 100% |
| Mistral | $1.54 | $4.84 | $0.154 | 1.05M | unknown | 100% |
| BaseTen | $2.10 | $6.60 | $0.210 | 1.05M | fp8 | 100% |
| BaseTen | $2.10 | $6.60 | $0.210 | 1.05M | fp8 | 100% |
| Baidu | $2.25 | $7.88 | $0.560 | 1.05M | fp4 | 100% |
| Alibaba | $2.31 | $7.26 | $0.462 | 1.00M | fp8 | 100% |
| Decart | $2.50 | $9.00 | $0.540 | 1.05M | fp4 | 100% |
8 of 34 hosts serve this model at 4-bit precision (InferenceNet, DeepInfra, CoreWeave, Inceptron, Nebius, Parasail, Baidu, Decart), at the same price as full-precision hosts. See why this matters.
What it costs to run
- Marketing emails
- $2.07
- per 1,000 · short brief in, one email out
- Support replies
- $2.00
- per 1,000 · ticket plus help-centre context
- Lead enrichment
- $3.32
- per 1,000 · company page in, fields out
- Long-form drafts
- $8.98
- per 1,000 · outline in, full draft out
- Questions on docs
- $8.02
- per 1,000 · one large document, many questions
- Agent / tool loop
- $7.26
- per 1,000 · long context reused each step
Priced on the median host. Change the workload on the leaderboard to see how the ranking moves.
Specification
- Context, most hosts
- 1.05M · 1,573 pages
- Context range across hosts
- 262K to 1.05M
- Max output
- 944K
- Cache discount
- 96% off cached input
- Input types
- text
- Tool use
- supported
- Reasoning
- optional
- Arena Elo
- 1268 across 8 categories
- Agentic Elo
- 1190
- AA Intelligence
- 33.7
Other models with the same host spread
Host pricing and precision read live from OpenRouter’s endpoint API, compiled 12 October 2026. Free to quote with attribution. See all findings.