Models & PricingModel Cards
May 28, 2026
GAPREMIERv26.05
SAGE 2.5 Celer
Hybrid model unifying instruct, reasoning, and coding in a single efficient model. 3x faster than comparable models at equivalent quality.
Speed
Performance
Context
32k
Price
$0.10 / $0.30 /M Tokens
Features
Features
Try it
import sagea
client = sagea.Client(api_key="your-api-key")
response = client.chat.completions.create(
model="sage-2-5-celer",
messages=[
{"role": "user", "content": "Write a Python function to sort a list"}
]
)
print(response.choices[0].message.content)Base URL: https://api.sagea.space/v1 — Auth: Authorization: Bearer $SAGEA_API_KEY
Specifications
| Parameters | 24B |
| Context Length | 32,000 tokens |
| Max Output | 4,096 tokens |
| Training Data | Up to May 2026 |
| License | Apache 2.0 |
Other Models
Capabilities
SAGE 2.5 Celer unifies instruct, reasoning, and coding in a single efficient 24B model. It delivers near-frontier quality at 3x the speed and a fraction of the cost.
- Unified architecture: one deployment handles chat, reasoning traces, and code completion.
- Efficient inference: sliding-window attention with 32k context, optimized for
POST https://api.sagea.space/v1/chat/completions. - Edge-ready: runs on consumer hardware and mobile-class GPUs for private deployments.
- Batch-friendly: half-price batching via
POST https://api.sagea.space/v1/batch.
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| Instruct following | 81.2 | Single-model setup |
| Code completion | 78.9 | 20+ languages |
| Latency (p50, 512 tokens) | 0.9s | 3x faster than 70B class |
Use cases
- High-throughput chatbots and support copilots
- Code autocomplete and review bots
- Cost-sensitive production APIs
Limitations
- 32k context — switch to SAGE 2.4 Actus for 128k repo-scale reasoning.
- For self-hosting with full weight access, see SAGE OSS.