Mouse-1 is the better buy: $0.10 per 1M input tokens against GLiDE's $0.15, zero data retention by default where GLiDE stores by default, and a slightly higher public benchmark average (84.9% vs 83.8%). Choose GLiDE if a single prompt needs up to 40k tokens, or if you want its adaptive thinking.
GLiDE is Fastino Labs' decision model, released at the end of September 2026 and served at /v1/systemone, the same request shape TypeSafe introduced. Fastino describes it as a thinking decision model that can spend more compute on hard questions. The Decision Index board scores it with thinking off. Mouse-1 does not generate thinking tokens; it reads probabilities directly.
At a glance
| Mouse-1 | Fastino GLiDE | |
|---|---|---|
| Price per 1M input tokens | $0.10 | $0.15 |
| Output tokens | Free | Free |
| Data retention | None by default | Stored by default; store: false or team ZDR |
| Public tool benchmarks (avg of 4) | 84.9% | 83.8% |
| Calibrated abstention | /v1/calibrations | Thresholds on your own data |
| Fine-tuning | /v1/tunings | Not for GLiDE |
| Largest prompt | 16,384 state tokens | 40,000 tokens (state + one question) |
| Images | 8 per request (beta) | 4 per request |
| Prompt-injection results | Published: 0% flipped | None published |
Public tool benchmarks
Average of BFCL, API-Bank, When2Call and PhishNChips (%) · Higher is better
| Mouse-1 | 84.9% |
|---|---|
| Fastino GLiDE | 83.8% |
Function calling, exact match (%) · Higher is better
| Fastino GLiDE | 97.8% |
|---|---|
| Mouse-1 | 97.6% |
Next API out of 53 (%) · Higher is better
| Fastino GLiDE | 85.0% |
|---|---|
| Mouse-1 | 83.7% |
Answer, call, ask or decline (%) · Higher is better
| Mouse-1 | 85.8% |
|---|---|
| Fastino GLiDE | 83.4% |
Phishing or legitimate email (%) · Higher is better
| Mouse-1 | 72.3% |
|---|---|
| Fastino GLiDE | 68.9% |
Mouse-1 leads on When2Call (85.8 vs 83.4) and PhishNChips (72.3 vs 68.9). GLiDE leads on BFCL (97.8 vs 97.6) and API-Bank (85 vs 83.7). The averages are close; the gaps that matter are on price and data handling.
Price and retention
At a third less per token, Mouse-1 costs $10 less per 200M input tokens than GLiDE.
The retention defaults point in opposite directions. Fastino stores /v1/systemone requests unless you opt out per request or per team, and its docs reserve opting out of platform training for enterprise customers. Mouse-1 never stores a request unless you set store: true to tune on it, and even then it is encrypted and deleted after 30 days.
How we measured
- GLiDE's scores are from the Jev Decision Index board (no-thinking configuration) as of October 20, 2026. Mouse-1 is not on the board; its scores come from our own run of the board's kit on the same rows.
- Product facts are from Fastino's documentation, checked October 20, 2026.
- We have not run GLiDE through our agent decision suites.
Sources
Checked October 20, 2026.
- Fastino: introducing GLiDE
- Fastino documentation
- Jev Decision Index board data
- Mouse-1 API reference and pricing
See also: Mouse-1 vs Perplexity Decider · Mouse-1 vs Jev · All comparisons
FAQ
Is Mouse-1 cheaper than Fastino GLiDE?
Yes. Mouse-1 costs $0.10 per 1M input tokens and GLiDE costs $0.15. Both charge nothing for output.
Does Fastino GLiDE store my requests?
By default, yes: Fastino's /v1/systemone stores requests unless you send store: false or turn on team-level zero data retention. Mouse-1 keeps nothing by default; storing a decision is an explicit opt-in for tuning.
Which scores higher on public benchmarks, Mouse-1 or GLiDE?
Mouse-1 averages 84.9% across BFCL, API-Bank, When2Call and PhishNChips against 83.8% for GLiDE (no-thinking) on the Jev Decision Index. Mouse-1 leads on When2Call and PhishNChips; GLiDE leads on BFCL and API-Bank.
Can I fine-tune GLiDE?
Fastino's docs say /v1/systemone does not accept fine-tuned deployments, and GLiDE is not among its trainable base models. Mouse-1 can be tuned on your own decision outcomes through /v1/tunings.