Gemma 3 27B
consumerAI-generated content marks
The provider reports that this model does not add embedded watermarks or provenance metadata to generated output.
- 27.2B
- 128k
- other
- 🇺🇸 USA
- Feb 2025
Scores
Score per dollar
1029 pts per $/M input
general_score (72) divided by cheapest input price ($0.07/M). Higher is better value. See live pricing.
Related models
Save your hardware and every model page answers the real question: will it run on your machine, and how fast?
Join free - save your rig →Run it locally
Per-quant memory needs and a static "can you run it?" reference - no rig entry required
The reference hardware
22 reference configs, drawn in-house. Scroll for more.
Can you run it? - reference rigs
| Rig | Q4_K_M | Q8_0 |
|---|---|---|
| NVIDIA Jetson Orin NX 16GB | no -> cloud | no -> cloud |
| Single GTX 1080 Ti (11GB) | offload | no -> cloud |
| 4x H100 80GB (320GB) | fast 367.2t/s | fast 237.2t/s |
| NVIDIA DGX Station 748GB | fast 219.2t/s | fast 141.6t/s |
| 8x RTX 3090 rack (192GB) | fast 205.2t/s | fast 132.6t/s |
| 4x RTX 5090 (128GB) | fast 196.4t/s | fast 126.9t/s |
| AMD Instinct MI300X (192GB) | fast 145.9t/s | fast 94.3t/s |
| 4x RTX 4090 (96GB) | fast 110.5t/s | fast 71.4t/s |
| 2x RTX 5090 (64GB) | fast 98.2t/s | fast 63.4t/s |
| 2x RTX 3090 (48GB) | fast 51.3t/s | fast 33.1t/s |
| Single RTX 5090 (32GB) | fast 49.1t/s | fast 31.7t/s |
| RTX PRO 6000 Blackwell (96GB) | fast 49.1t/s | fast 31.7t/s |
| Mac Studio M4 Ultra 192GB | fast 32.6t/s | fast 21.1t/s |
| Mac Studio M4 Ultra 512GB | fast 32.6t/s | fast 21.1t/s |
| Single RTX 4090 (24GB) | fast 27.6t/s | offload |
| MacBook Pro M5 Max 128GB | ok 18.4t/s | ok 11.9t/s |
| Dual EPYC 9004 + 768GB DDR5-4800 | ok 12.6t/s | ok 8.2t/s |
| DGX Spark 128GB unified | slow 7.5t/s | slow 4.8t/s |
| Ryzen AI Max+ 395 128GB | slow 7.0t/s | slow 4.5t/s |
| Jetson AGX Orin 64GB | slow 5.6t/s | slow 3.6t/s |
| Epyc + 512GB DDR4-3200 + 2x RTX 3090 | slow 5.6t/s | slow 3.6t/s |
| Epyc + 512GB DDR4-2400 + 2x RTX 3090 | slow 4.2t/s | slow 2.7t/s |
Fit tiers use the same will-it-run logic as the rig finder. For comfortable fits, the badge reflects decode speed: fast >=20 t/s, ok 8-20 t/s, slow <8 t/s. t/s is a bandwidth estimate, not a measured benchmark.
Download options
Or run it in the cloud
Live per-provider pricing, throughput and uptime - refreshed about 1 hour ago via OpenRouter. Click a column to sort.
| Provider | Type | Input $/M | Output $/M | Cache $/M | Tok/s | Latency | Uptime | Value |
|---|---|---|---|---|---|---|---|---|
|
NextBit
|
API | 0.07 | 0.28 | 0.040 | - | - | 100.00% | best uptime |
|
Parasail
|
API | 0.08 | 0.45 | 0.040 | - | - | 96.67% | |
|
DeepInfra
risky
|
API | 0.08 | 0.16 | 0.040 | - | - | 93.33% | |
|
Novita
risky
|
API | 0.12 | 0.20 | 0.040 | - | - | 93.33% | |
|
Nebius
risky
|
API | 0.10 | 0.30 | 0.040 | - | - | 90.00% |
Default order: throughput among 95%+ uptime providers, then latency; subscriptions last. Sort by any column. Subscription rows show $/mo in the Value column - per-token columns are "-". Affiliate links are marked sponsored / nofollow. Confirm current pricing on the provider's site before committing.
Detailed API pricing page + JSON endpoint →
See who runs Google in production →
Inference cost over time
Data accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.