Ho Boloka Prompt ka Cache ho Elang Hloko Provider

Superdav AI Agent v1.12.0 e hlahisa ho boloka prompt ka cache ho elang hloko provider, ho ntlafatsang ditjeho tsa API le tieho ka ho boloka prompts ka cache ho pholletsa le providers ba fapaneng ba LLM. Provider ka nngwe e na le mekgwa le ditlhophiso tse fapaneng tsa caching.

Kakaretso

Ho boloka prompt ka cache ho o dumella ho:

• Boloka prompts tse kgolo, tse sebediswang kgafetsa ka cache
• Fokotsa ditjeho tsa API ka ho qoba tshebetso e iphetang e sa hlokahaleng
• Ntlafatsa tieho bakeng sa dikopo tse bolokilweng ka cache
• Laola potoloho ya bophelo ba cache ka ho hlaka

Providers ba fapaneng ba kenya caching tshebetsong ka ditsela tse fapaneng:

• Google Gemini: cachedContents API
• Azure OpenAI: Ho boloka prompt ka cache ka TTL
• OpenRouter: Caching e ikgethileng ho provider
• Vertex Anthropic: Ho boloka prompt ka cache ka taolo ya cache

Google Gemini: cachedContents API

Google Gemini e fana ka taolo e hlakileng ya cache ka cachedContents API.

Tlhophiso

$config = [
    'provider' => 'google-gemini',
    'model' => 'gemini-2.0-flash',
    'caching' => [
        'enabled' => true,
        'ttl' => 3600, // 1 hour in seconds
        'max_tokens' => 1000000, // Max tokens to cache
    ],
];

Ho Theha Prompt e Bolokilweng ka Cache

use Superdav\AI\Providers\GoogleGemini;

$gemini = new GoogleGemini( $config );

$cached_content = $gemini->create_cached_content(
    [
        'system_prompt' => 'You are a helpful assistant...',
        'context' => 'Large context document...',
        'ttl' => 3600,
    ]
);

// Returns: ['cache_id' => 'abc123', 'expires_at' => timestamp]

Ho Sebedisa Prompt e Bolokilweng ka Cache

$response = $gemini->generate(
    [
        'cache_id' => 'abc123',
        'prompt' => 'User question here',
    ]
);

Potoloho ya Bophelo ba Cache

// List cached contents
$caches = $gemini->list_cached_contents();

// Get cache details
$cache = $gemini->get_cached_content( 'abc123' );

// Extend cache TTL
$gemini->update_cached_content(
    'abc123',
    ['ttl' => 7200] // Extend to 2 hours
);

// Delete cache
$gemini->delete_cached_content( 'abc123' );

Mekgwa e Metle ka ho Fetisisa bakeng sa Gemini

• Beha TTL e loketseng: Lekalekanya poloko ya ditjeho le ho tsofala ha cache
• Boloka system prompts ka cache: Sebedisa system prompt e tshwanang hape dikopong tse fapaneng
• Lekola tshebediso ya cache: Latela hore na ke di-cache dife tse sebediswang haholo
• Hlakola di-cache tse fedileng nako: Hlakola di-cache tse sa sebedisweng nako le nako

Azure OpenAI: Ho Boloka Prompt ka Cache

Azure OpenAI e tshehetsa ho boloka prompt ka cache ka taolo ya TTL e iketsang.

Tlhophiso

$config = [
    'provider' => 'azure-openai',
    'model' => 'gpt-4-turbo',
    'api_version' => '2024-08-01-preview',
    'caching' => [
        'enabled' => true,
        'cache_control' => 'max_age=3600',
    ],
];

Ho Dumella Caching

use Superdav\AI\Providers\AzureOpenAI;

$azure = new AzureOpenAI( $config );

$response = $azure->generate(
    [
        'system_prompt' => 'You are a helpful assistant...',
        'context' => 'Large context document...',
        'prompt' => 'User question here',
        'cache_control' => 'max_age=3600',
    ]
);

// Response includes cache usage:
// [
//     'content' => '...',
//     'cache_creation_input_tokens' => 1000,
//     'cache_read_input_tokens' => 500,
// ]

Dihlooho tsa Cache

Azure OpenAI e sebedisa dihlooho tsa HTTP bakeng sa taolo ya cache:

Cache-Control: max_age=3600

Diboleng tse tshehetswang:

• max_age=<seconds>: Boloka ka cache bakeng sa nako e boletsweng
• no_cache: O se ke wa boloka kopo ena ka cache
• no_store: O se ke wa boloka ka cache mme o se ke wa sebedisa hape

Ho Lekola Tshebediso ya Cache

$response = $azure->generate( [...] );

$cache_tokens = $response['cache_creation_input_tokens'] ?? 0;
$cache_hits = $response['cache_read_input_tokens'] ?? 0;

echo "Cache creation: $cache_tokens tokens\n";
echo "Cache hits: $cache_hits tokens\n";

Mekgwa e Metle ka ho Fetisisa bakeng sa Azure OpenAI

• Sebedisa prompts tse tsitsitseng: Prompts tse tshwanang di una molemo ho caching
• Beha TTL e utlwahalang: Lekalekanya ditjeho le botjha
• Lekola metrics tsa cache: Latela ho thehwa ha cache kgahlanong le hits
• Kopanya dikopo tse tshwanang: Hlophisa dikopo ka dihlopha ho eketsa cache hits

OpenRouter: Caching e Ikgethileng ho Provider

OpenRouter e tshehetsa caching ka providers ba ka tlase (OpenAI, Anthropic, jj.).

Tlhophiso

$config = [
    'provider' => 'openrouter',
    'model' => 'openai/gpt-4-turbo',
    'caching' => [
        'enabled' => true,
        'provider_cache' => 'openai', // Use OpenAI's caching
    ],
];

Ho Sebedisa OpenRouter Caching

use Superdav\AI\Providers\OpenRouter;

$router = new OpenRouter( $config );

$response = $router->generate(
    [
        'system_prompt' => 'You are a helpful assistant...',
        'context' => 'Large context document...',
        'prompt' => 'User question here',
        'cache_control' => 'max_age=3600',
    ]
);

Dikgetho tse Ikgethileng ho Provider

Providers ba fapaneng ba na le mekgwa e fapaneng ya caching:

// OpenAI-compatible caching
$response = $router->generate(
    [
        'model' => 'openai/gpt-4-turbo',
        'cache_control' => 'max_age=3600',
    ]
);

// Anthropic-compatible caching
$response = $router->generate(
    [
        'model' => 'anthropic/claude-3-opus',
        'cache_control' => [
            'type' => 'ephemeral',
            'max_tokens' => 1000000,
        ],
    ]
);

Mekgwa e Metle ka ho Fetisisa bakeng sa OpenRouter

• Tseba caching ya provider wa hao: Provider ka nngwe e na le mekgwa e fapaneng
• Leka boitshwaro ba caching: Netefatsa hore caching e sebetsa le provider eo o e kgethileng
• Lekola ditjeho: Latela poloko e tswang ho caching
• Sebedisa models tse tsitsitseng: Ho fetola models ho senya cache hits

Vertex Anthropic: Ho Boloka Prompt ka Cache ka Taolo ya Cache

Vertex Anthropic (Google Cloud) e tshehetsa ho boloka prompt ka cache ka taolo e hlakileng ya cache.

Tlhophiso

$config = [
    'provider' => 'vertex-anthropic',
    'model' => 'claude-3-opus',
    'project_id' => 'your-gcp-project',
    'region' => 'us-central1',
    'caching' => [
        'enabled' => true,
        'cache_control' => [
            'type' => 'ephemeral',
            'max_tokens' => 1000000,
        ],
    ],
];

Ho Sebelisa Vertex Anthropic Caching

use Superdav\AI\Providers\VertexAnthropic;

$vertex = new VertexAnthropic( $config );

$response = $vertex->generate(
    [
        'system_prompt' => 'You are a helpful assistant...',
        'context' => 'Large context document...',
        'prompt' => 'User question here',
        'cache_control' => [
            'type' => 'ephemeral',
            'max_tokens' => 1000000,
        ],
    ]
);

// Response includes cache metrics:
// [
//     'content' => '...',
//     'usage' => [
//         'input_tokens' => 1000,
//         'cache_creation_input_tokens' => 500,
//         'cache_read_input_tokens' => 300,
//     ],
// ]

Mefuta ea Taolo ea Cache

• ephemeral: Cache bakeng sa nako ea kopo (ea kamehla)
• persistent: Cache ho pholletsa le likopo tse ngata (haeba e tšehetsoa)

Ho Hlokomela Tšebeliso ea Cache

$response = $vertex->generate( [...] );

$usage = $response['usage'];
$cache_created = $usage['cache_creation_input_tokens'] ?? 0;
$cache_read = $usage['cache_read_input_tokens'] ?? 0;

echo "Cache created: $cache_created tokens\n";
echo "Cache read: $cache_read tokens\n";

Mekhoa e Metle bakeng sa Vertex Anthropic

• Sebelisa caching ea ephemeral: E loketse caching ea seboka se le seng
• Beha max_tokens ka nepo: Lekalekanya boholo ba cache le litšenyehelo
• Hlokomela metrics ea cache: Latela katleho ea cache
• Lekola ka mosebetsi oa hau: Netefatsa hore caching e tsoela nyeoe ea hau ea tšebeliso molemo

Leano la Caching Pakeng tsa Bafani

Tlhophiso e Kopaneng

$config = [
    'caching' => [
        'enabled' => true,
        'default_ttl' => 3600,
        'providers' => [
            'google-gemini' => [
                'ttl' => 3600,
                'max_tokens' => 1000000,
            ],
            'azure-openai' => [
                'cache_control' => 'max_age=3600',
            ],
            'vertex-anthropic' => [
                'cache_control' => [
                    'type' => 'ephemeral',
                    'max_tokens' => 1000000,
                ],
            ],
        ],
    ],
];

Ho Lemohwa ha Mofani

$provider = $config['provider'];

$cache_config = $config['caching']['providers'][ $provider ]
    ?? $config['caching'];

// Use provider-specific caching configuration

Leano la Fallback

try {
    // Try caching with primary provider
    $response = $primary_provider->generate( $request );
} catch ( CacheException $e ) {
    // Fall back to non-cached request
    $response = $primary_provider->generate(
        array_merge( $request, ['cache_control' => 'no_cache'] )
    );
}

Ntlafatso ea Litšenyehelo

Bala Poloko

$cache_created_tokens = $response['cache_creation_input_tokens'] ?? 0;
$cache_read_tokens = $response['cache_read_input_tokens'] ?? 0;
$regular_tokens = $response['input_tokens'] ?? 0;

// Typical pricing (varies by provider):
$cache_creation_cost = $cache_created_tokens * 0.00001; // 10x cheaper
$cache_read_cost = $cache_read_tokens * 0.000001; // 100x cheaper
$regular_cost = $regular_tokens * 0.00001;

$total_cost = $cache_creation_cost + $cache_read_cost + $regular_cost;
$savings = ($regular_tokens * 0.00001) - $total_cost;

echo "Estimated savings: \$$savings\n";

Malebela a Ntlafatso

• Cache system prompts tse kholo: Poloko e kholo ka ho fetisisa ea litšenyehelo
• Sebelisa moelelo hape: Cache litokomane tsa moelelo tse sebelisoang khafetsa
• Likopo tsa batch: Kopanya likopo tse tšoanang ho eketsa cache hits
• Hlokomela katleho ea cache: Latela poloko ea sebele
• Fetola TTL: Lekalekanya litšenyehelo le bocha

Ho Rarolla Mathata

Cache ha e sebelisoe

• Netefatsa hore caching e nolofalitsoe tlhophisong
• Hlahloba hore prompts lia tšoana (caching e hloka ho tšoana ka ho feletseng)
• Netefatsa hore cache ha e so fele
• Hlahloba meeli ea cache e ikhethileng ho mofani

Ho theha cache hoa hloleha

• Netefatsa hore boholo ba cache bo ka har’a meeli ea mofani
• Hlahloba hore syntax ea taolo ea cache e nepahetse
• Netefatsa hore mofani o tšehetsa caching bakeng sa model ea hau
• Hlahloba litokomane tsa mofani bakeng sa meeli

Litšenyehelo tse sa lebelloang

• Hlokomela ho thehoa ha cache ha ho bapisoa le tokens tsa ho bala cache
• Netefatsa hore cache e hlile e sebelisoa
• Hlahloba cache misses tse bakoang ke liphapang tsa prompt
• Nahana ka ho fetola TTL kapa leano la cache

Papiso ea Bafani

Tšobotsi	Gemini	Azure OpenAI	OpenRouter	Vertex Anthropic
Cache API	cachedContents	HTTP headers	E ikhethileng ho mofani	Taolo ea cache
Taolo ea TTL	E hlakileng	Ka headers	E itšetlehile ka mofani	Ephemeral/persistent
Boholo bo boholo ba cache	1M tokens	Bo itšetlehile ka mofani	Bo itšetlehile ka mofani	1M tokens
Phokotso ea litšenyehelo	90%	90%	E itšetlehile ka mofani	90%
Ho hlokomela	Ka botlalo	Ka metrics	Ho itšetlehile ka mofani	Ka usage

Mehato e Latelang

1. Khetha mofani oa hau: Khetha ho latela litlhoko tsa hau
2. Hlophisa caching: Beha caching e ikhethileng ho mofani
3. Lekola caching: Netefatsa hore e sebetsa le prompts tsa hau
4. Hlokomela tšebeliso: Latela cache hits le poloko ea litšenyehelo
5. Ntlafatsa: Fetola TTL le leano la cache ho latela liphetho