v0.3.4

fix: reinstated rust-tls to reqwest. Updated current model options.  …

v0.3.3

• make error handling report response body
• support custom models (so when new ones are out, they can be used)
• allow setting response_schema
• safety_ratings is optional in response
• upgrade reqwest and make default features optional

v0.3.2

V.0.3.2 (#39)

- `gcp_auth` version bump
- PR #38

v3.0.0

Added beta feature flag.

• New model: gemini-1.5-pro-latest
• system instructions
• json_mode

v0.2.5

chore: version 0.2.5

v0.2.4

chore: bump to v0.2.4

Signed-off-by: Mick Clarke <avastmick@users.noreply.github.com>

v0.2.3

Added character counts to allow for cost tracking, as per the Vertex AI pricing.

Refactored out the Vertex AI-specific code into own module.

v0.2.2

generateContent is now viable for Vertex AI endpoint.

v0.2.1

Streamed response, plus fixes (#22)

Fixes for:

- Issue #12 
- Issue #19.

Added feature:

- `countToken` - see [Gemini
API::countTokens](https://ai.google.dev/tutorials/rest_quickstart#count_tokens).